A CRITICAL ANALYSIS OF THE CURRENT APPROACH OF THE
COURTS AND ACADEMICS TO THE PROBLEM OF EVIDENTIAL
UNCERTAINTY IN CAUSATION IN TORT LAW
By
Gemma Turton
A thesis submitted to the
University of Birmingham
for the degree of
DOCTOR OF PHILOSOPHY
School of Law
College of Arts, Humanities, and Law
University of Birmingham
October 2012
University of Birmingham Research Archive
e-theses repository
This unpublished thesis/dissertation is copyright of the author and/or third
parties. The intellectual property rights of the author or third parties in respect
of this work are as defined by The Copyright Designs and Patents Act 1988 or
as modified by any successor legislation.
Any use made of information contained in this thesis/dissertation must be in
accordance with that legislation and must be properly acknowledged. Further
distribution or reproduction in any format is prohibited without the permission
of the copyright holder.
Abstract
The primary aim of this thesis is to identify a coherent legal response to the particular causal
problem of the ‘evidentiary gap’. In order to do this, it is necessary to understand how the
‘evidentiary gap’ relates to causation in negligence more generally, so the thesis addresses both
the nature and function of the tort of negligence as well as the role played by causation within
that tort. It argues that negligence is best understood as a system of corrective justice-based
interpersonal responsibility. In this account, causation has a vital role so the test of causation
must be philosophically sound. Causation, however, also occupies only a limited role so analysis
must draw fully on the doctrines of damage and breach which bracket the causation inquiry, as
well as notions of quantification of loss. The NESS test for causation is shown to be preferable
to the but-for test because it is conceptually more adequate and therefore able to address causal
problems that the but-for test cannot. This thesis rejects claims for proportionate recovery based
on the notion of loss of a chance of avoiding physical harm in medical negligence, but proposes
limited recovery for loss of a chance as an independent form of damage arising because of
unique considerations of interpersonal responsibility in the doctor-patient relationship in cases of
misdiagnosis/mistreatment of existing illness. It is argued that the Fairchild test of material
contribution to risk of harm in cases of evidentiary gap is not consistent with corrective justice,
and that this cannot be resolved by reconceptualising the gist of the action as the risk of harm.
The Fairchild exception lacks coherence because of its instrumentalist basis, so should not be
applied outside of the mesothelioma context.
i
Acknowledgments
I would like to thank my supervisor, Claire McIvor, for all her help and support throughout the
time it has taken to research and write this thesis. She has not only guided me academically but
has also developed my confidence and self-belief. I owe her an enormous debt of gratitude. I
would also like to thank Sarah Green for inspiring my interest in causation as an undergraduate
student. I am grateful to the University of Birmingham for giving me the postgraduate teaching
assistantship that made this financially possible, and to the University of Leicester for giving me
a period of study leave. I have been fortunate to present papers on aspects of this research, so I
would like to thank the School of Law and the Institute of Medical Law at the University of
Birmingham, the School of Law at the University of Leicester, the European Association of
Health Law, and the Obligations VI conference for allowing me this opportunity. A number of
people have read work on aspects of the thesis, and I would particularly like to thank James Lee,
John Hartshorne, and Jose Miola for this.
I would like to thank a number of my friends who have been incredibly supportive (and patient),
especially Martin George, Laura Brampton, Sophie Boyron, Sally Cunningham, Daniel
Attenborough, and Paul O’Connell. I owe great thanks to my parents for their encouragement.
Most of all, I would like to thank my sister, Samantha, for her love and encouragement
throughout the years. She is an inspiration in herself and none of this would have been possible
without her.
ii
Table of Contents
Introduction ..........................................................................................................................1
Chapter One: Theoretical and doctrinal framework ........................................................... 6
1.
The theoretical basis of negligence................................................................................................... 7
1.1 The Aristotelian conception of corrective justice .................................................................. 11
1.1.1 Justice as a whole ............................................................................................................. 11
1.1.2 Justice as a part: corrective and distributive justice .................................................... 12
1.1.2.1 Distributive justice .................................................................................................. 13
1.1.2.2 Corrective justice..................................................................................................... 14
1.2 The moral basis of corrective justice: Aristotelian corrective justice and Kantian Right 17
1.2.1 Kantian Right ................................................................................................................... 19
1.3 Incoherence of consequentialist theories .............................................................................. 22
1.4
Moving beyond formalism to the legal system - combining Kantian Right and societal
morality…. ......................................................................................................................................... 26
1.5 The place of causation in corrective justice-based interpersonal responsibility .............. 32
1.5.1 The pivotal place of causation in corrective justice.................................................... 32
1.5.2 The limited role of causation ......................................................................................... 35
2.
The doctrinal framework ................................................................................................................. 38
2.1 The duty of care ........................................................................................................................ 39
2.1.1 Duty as the ‘central organising concept’ ...................................................................... 39
2.1.2 Identifying issues of interpersonal responsibility in the duty doctrine .................... 42
2.2 Actionable damage/the gist of the negligence action.......................................................... 48
2.2.1 Damage ............................................................................................................................. 49
2.2.2 Quantification .................................................................................................................. 49
2.3 Breach of duty ........................................................................................................................... 52
Chapter Two: Identifying the Proper Function of Causation ........................................... 55
1.
The demands on causation .............................................................................................................. 57
iii
1.1 The limited role of causation in negligence .......................................................................... 57
1.2 The robustness of the philosophical account ....................................................................... 61
2.
The philosophical account of causation ........................................................................................ 63
2.1 Hume: contiguity, priority, constant conjunction and necessity ........................................ 63
2.2 Mill: the addition of ‘plurality’ and ‘complexity’................................................................... 65
2.2.1 Complexity of causes ...................................................................................................... 66
2.2.2 Plurality of causes ............................................................................................................ 68
3.
Tests for causation ............................................................................................................................ 70
3.1 Complexity and plurality of causes as necessary and sufficient conditions...................... 70
3.2 But-for and NESS in simple cases ......................................................................................... 72
3.3 Over-determined causation ..................................................................................................... 74
3.3.1 But-for, NESS and pre-emption ................................................................................... 75
3.3.2 But-for, NESS and duplication ..................................................................................... 76
4.
Overcoming potential problems with NESS ................................................................................ 79
4.1 Defining key causal terms ........................................................................................................ 80
4.2 The problem of asymmetrical duplication ............................................................................ 82
4.2.1 The solution to asymmetrical duplication.................................................................... 83
4.2.2 Causation of divisible damage ....................................................................................... 88
4.2.2.1 Divisible damage and interaction between causes ............................................. 91
5.
Using NESS to overcome common problems with exceptional legal tests ............................. 92
5.1 Bonnington Castings v Wardlaw .................................................................................................... 94
5.1.1 The scope of the test ...................................................................................................... 96
5.1.1.1 Divisibility of damage............................................................................................. 96
5.1.1.2 ‘Cumulative’ causal process ................................................................................... 98
5.1.2 Analysis of the Wardlaw test........................................................................................... 99
5.2 Practical application of the NESS test ................................................................................. 101
5.2.1 The question of causation in Bailey ............................................................................. 101
5.2.2 The question of ‘apportionment’ ................................................................................ 103
iv
Chapter Three: Loss of a chance ...................................................................................... 107
1.
Analysis of Hotson v East Berkshire Health Authority .................................................................... 108
1.1 The facts ................................................................................................................................... 109
1.2 Distinguishing damage, causation, and quantification ...................................................... 111
1.2.1 The ‘hook’ argument..................................................................................................... 112
1.2.2 The vicissitudes principle ............................................................................................. 114
1.3 The contract/tort distinction ................................................................................................ 116
2.
Analysis of Gregg v Scott .............................................................................................................. 118
2.1 Pain and suffering ................................................................................................................... 119
2.2 Loss of life expectancy ........................................................................................................... 121
3.
Statistics, chance, and the balance of probabilities standard of proof .................................... 123
3.1 What must the claimant prove? What is the problem of proof? ..................................... 125
3.1.1 Damage ........................................................................................................................... 126
3.1.2 Causation ........................................................................................................................ 126
3.2 The balance of probabilities standard of proof .................................................................. 129
3.3 Proof in Hotson ........................................................................................................................ 134
3.4 Proof in Gregg .......................................................................................................................... 136
3.4.1 Reliability of the evidence ............................................................................................ 138
3.4.2 Personalising the evidence ........................................................................................... 139
4.
‘Loss of a chance’ as damage......................................................................................................... 141
4.1 Rejecting ‘chance’ conceived as a proportion of the physical outcome ......................... 142
4.2 Epistemological chance as an independent form of damage ........................................... 145
4.2.1 Post-diagnosis factors that determine the outcome ................................................. 146
4.2.2 The grey-areas on the deterministic-indeterministic spectrum ............................... 148
4.2.3 When does a patient have a ‘chance’? Chance vs. risk. ............................................ 152
4.2.3.1When does a patient have a meaningful chance? .............................................. 154
4.2.4 ‘Loss’ of a chance .......................................................................................................... 158
4.2.5 Valuing the chance ........................................................................................................ 160
v
Chapter Four: The evidentiary gap .................................................................................. 164
1.
Defining the evidentiary gap ......................................................................................................... 166
1.1 The evidentiary gap relating to dermatitis: McGhee v National Coal Board ......................... 167
1.2 The evidentiary gap relating to mesothelioma: Fairchild v Glenhaven Funeral Services ........ 168
1.3 Analysis of the evidentiary gap ............................................................................................... 171
1.3.1 NESS and identifying the problem: locating the evidentiary gap .......................... 172
1.3.2 NESS: separating the conceptual from the evidential aspects of the legal solution
………………………………………………………………………………………….174
2.
Evidential solutions: the inferential interpretation of McGhee .................................................. 177
2.1. The inferential approach to McGhee ...................................................................................... 178
2.1.1 Scientific and legal standards of proof ....................................................................... 180
2.1.1.1 The place of intuition in the balance of probabilities ...................................... 183
2.1.1.2 Evidence of the evidentiary gap ......................................................................... 184
2.2 The ‘doubling of the risk’ approach to proof on the balance of probabilities ................ 187
2.2.1 Doubling of the risk ...................................................................................................... 188
2.2.2 Closing the evidentiary gap .......................................................................................... 191
3. Risk of harm as the gist of the negligence action: Barker v Corus (UK) plc .................................. 194
3.1 The decision in Barker v Corus ................................................................................................. 195
3.2 Analysis of the ‘risk as gist’ approach .................................................................................... 199
3.2.1 The inconsistency of the physical harm requirement. ............................................. 201
3.2.2 Physical harm and risk: the shift in perspectives ...................................................... 202
3.2.2.1 What is ‘risk’? ......................................................................................................... 203
3.2.2.2 Can risk constitute damage? ................................................................................ 205
3.2.2.2.1 Calculating risk and probability .................................................................. 206
3.2.2.2.2 ‘Increase in risk’ or ‘contribution to risk’? ................................................ 208
3.2.3 Confusion in the ‘apportionment’ issue ..................................................................... 210
3.3 Can risk constitute damage in corrective justice? ................................................................ 211
vi
3.3.1 Objective risk: the evidentiary gap prevents proof that the defendant created an
objective risk... ..........…………………………………………………………………213
3.3.2 Subjective/epistemic risk: lacks the moral significance to be damage .................. 214
3.3.2.1 The difficulty of explaining why epistemic risk deserves compensation. ..... 215
3.3.2.2 Epistemic risk: the conflation of breach and damage ..................................... 218
4. Theoretical approaches to the ‘evidentiary gap’ ............................................................................. 220
4.1 Dispelling possible misconceptions ....................................................................................... 220
4.1.1 McGhee: the defendant definitely caused the loss ...................................................... 221
4.1.2 Fairchild: the ‘innocent claimant’ argument ................................................................ 222
4.2 Where next? ............................................................................................................................... 226
4.2.1 Mesothelioma................................................................................................................. 226
4.2.2 Other diseases involving an evidentiary gap ............................................................. 227
4.2.2.1 The single harmful agent rule: Wilsher v Essex Area Health Authority ............. 228
4.2.2.2 Limiting the Fairchild exception to mesothelioma ............................................ 232
Conclusion ........................................................................................................................ 239
vii
Table of Cases
Allen v British Rail Engineering Ltd (BREL) [2001] EWCA Civ 242, [2001] ICR 942
Bailey v Ministry of Defence and another [2007] EWHC 2913 (QB); [2008] EWCA Civ 883, [2009] 1
WLR 1052
Baker v Willoughby [1970] AC 467 (HL)
Barker v Corus (UK) plc [2006] UKHL 20, [2006] 2 AC 572
Barnett v Chelsea and Kensington Hospital Management Committee [1968] 2 WLR 422 (QBD)
Bolitho v City of Hackney Health Authority [1998] AC 232 (HL)
Bonnington Castings v Wardlaw [1956] AC 613 (HL)
Cartledge v E Jopling & Sons Ltd [1963] AC 758 (HL)
Chaplin v Hicks [1911] 2 KB 786 (CA)
Chester v Afshar [2004] UKHL 41, [2005] 1 AC 134
Dean v Keate (1811) 3 Camp. 4 (170 ER 1286)
Dickins v O2 plc [2008] EWCA Civ 1144, [2009] IRLR 58
Dillon v Twin State Gas & Electric Co, 85 NH 449, 169 A 111 (1932)
Dingley v The Chief Constable, Strathclyde Police 1998 SC 548
Durham v BAI (Employers’ Liability Insurance “Trigger” Litigation) [2012] UKSC 14, [2012] 1 WLR
867
Jobling v Associated Dairies [1982] AC 794 (HL)
Fairchild v Glenhaven Funeral Services (t/a GH Dovener & Son) [2001] EWCA Civ 1881, [2002] 1
WLR 1052; [2002] UKHL 22, [2003] 1 AC 32
Gouldsmith v Mid-Staffordshire General Hospitals Trust [2007] EWCA Civ 397, [2007] LS Law Medical
363
Gregg v Scott [2002] EWCA Civ 1471, [2003] Lloyd's Rep Med 105; [2005] UKHL 2, [2005] 2 AC
176
Gwilliam v West Herts NHS Trust [2002] EWCA Civ 1041, [2003] QB 443
Hatton v Sutherland [2002] EWCA Civ 76, [2002] 2 All ER 1
Hinz v Berry [1970] 2 QB 40 (CA)
Holtby v Brigham & Cowan (Hull) Ltd [2000] 3 All ER 421 (CA)
Hotson v Fitzgerald and others [1985] 1 WLR 1036 (QBD)
Hotson v East Berkshire Health Authority [1987] 2 WLR 287 (CA); [1987] AC 750 (HL)
Jobling v Associated Dairies Ltd [1982] AC 794 (HL)
Lim Poh Choo v Camden & Islington AHA [1980] AC 174 (HL)
viii
McGhee v National Coal Board 1972 SLT (notes) 61 (Court of Session, Scotland); [1973] 1 WLR 1
(HL (Sc.))
McWilliams v Sir William Arrol Co Ltd [1962] 1 WLR 295 (HL)
Nicholson v Atlas Steel Foundry & Engineering Co Ltd [1957] 1 WLR 613 (HL (Sc))
Overseas Tankship (UK) Ltd v Morts Dock and Engineering Co Ltd (The Wagon Mound (No1)) [1961] AC
388 (PC)
Performance Cars v Abraham [1962] 1 QB 33 (CA)
Rahman v Arearose Ltd [2001] QB 351 (CA)
Re Polemis and Furness Withy Co [1921] 3 KB 560
Reeves v Commissioner of Police of the Metropolis [2000] 1 AC 360 (HL)
Rothwell v Chemical and Insulating Co Ltd and another [2007] UKHL 39, [2007] 3 WLR 876
Sienkiewicz v Greif (UK) Ltd [2011] UKSC 10; [2011] 2 WLR 523
Sindell v Abbott Laboratories 26 Cal. 3d 588, 607 P 2d 924 (1980)
Smith v Leech Brain & Co [1962] 2 QB 405 (QBD)
South Australia Asset Management Corp (“SAAMCo”) v York Montague Ltd [1997] AC 191 (HL)
Thompson v Smiths Shiprepairers (North Shields) Ltd [1984] QB 405 (QBD)
Williams v University of Birmingham [2011] EWCA Civ 1242, [2012] PIQR P4
Wilsher v Essex AHA [1987] 1 QB 730 (CA); [1988] AC 1074 (HL)
ix
Introduction
It has been said that ‘[t]o insist on a causal connection between conduct ensures that in general
we impose liability only on those who, by intervening in the world, have changed the course of
events for the worse’.1 The factual causation requirement is therefore an essential ingredient of
negligence liability, yet in recent years the approaches of courts and academics to the doctrine of
causation have become increasingly complicated and confused.
The primary aim of this thesis is to identify a coherent legal response to the particular causal
problem of the ‘evidentiary gap’. The ‘evidentiary gap’ problem arises where a claimant has been
exposed to a harmful agent from a number of sources, including the defendant’s negligence, and
a gap in scientific understanding of the aetiology of the disease prevents proof that any particular
source of the harmful agent was actually a cause of the harm. This is a rare problem that arises
because medical science cannot explain how the disease occurs so it represents an
insurmountable problem of proof for the claimant. This was first addressed in McGhee2 and
Fairchild,3 where the House of Lords relaxed the causation requirement and allowed the claimants
to recover on the basis of proof that the defendant’s negligence had ‘materially increased the risk
of harm’. In order to identify a coherent solution to the ‘evidentiary gap’ it will first be necessary
to situate the issue within the broader context of causation in negligence more generally. This
will involve addressing both the nature and function of the tort of negligence as well as the role
played by causation within that tort.
The thesis will therefore begin with an analysis of the theoretical basis of negligence. In Chapter
One, it will be argued that negligence is best understood in terms of corrective justice-based
interpersonal responsibility because this provides an account of negligent that prioritises
coherence and morality in the law. After explaining that causation occupies a pivotal yet limited
Tony Honoré, Responsibility and Fault (Hart Publishing 1999) 120.
McGhee v National Coal Board [1973] 1 WLR 1 (HL (Sc.)).
3 Fairchild v Glenhaven Funeral Services (t/a GH Dovener & Son) [2002] UKHL 22, [2003] 1 AC 32.
1
2
1
role within interpersonal responsibility, the chapter will turn to the doctrinal framework of
negligence in order to show how this role for causation is to be translated into practice. This
section will illustrate the interrelationship between the various negligence doctrines which each
fulfil a distinct purpose yet join together to form an integrated whole. Subsequent analysis of
caselaw on causation will draw on this to show that some problems which are currently treated
as being causal are not truly causal and that they should be addressed through other negligence
doctrines. The analysis of the doctrinal framework will also highlight the centrality of the duty
concept, and will argue that the issues of interpersonal responsibility identified at the duty of care
stage of the negligence inquiry should shape the application of the remaining doctrines. Later
chapters will draw on this idea when addressing problems of causation.
Chapter Two then turns to the concept of causation and the demands that the law places on a
test for causation. Since causation is pivotal to interpersonal responsibility it is important that the
legal approach to causation is premised on a philosophically sound account of the concept of
causation. In this chapter it will be argued that Wright’s NESS test more accurately translates the
philosophical account of causation into a workable test than the but-for test that currently
dominates judicial approaches to causation. In the majority of negligence cases where causation
is relatively straightforward, the NESS test is no more complicated to apply than the but-for test,
but in the small number of cases where the causal problem is more complex the NESS test is
able to identify causes where the but-for test fails.
It will be demonstrated that, under NESS, it is essential to define the damage that forms the gist
of the negligence action, and that the causal problem will vary depending on whether the
harmful outcome is ‘divisible’ i.e. dose-related, or ‘indivisible’ (‘all-or-nothing’). Chapter Two will
also evaluate the Wardlaw test of ‘material contribution to harm’.4 It is currently unclear when
and why this test applies, and therefore it is unclear whether it is an exception to the but-for test
and/or to the causation requirement. Using the NESS analysis it will be shown that the Wardlaw
4
Bonnington Castings v Wardlaw [1956] AC 613 (HL).
2
test seems to address a wide range of causal problems: where damage is divisible it equates to an
application of the but-for test to a portion of the harm, where damage is indivisible it is often an
exception to the but-for test but not an exception to the causation requirement. In these
circumstances the Wardlaw test compensates for the conceptual inadequacies of the but-for test.
This has important consequences for the approach to evidentiary uncertainty. The NESS test
will enable us to see that cases where the Wardlaw test is applied are not really problematic, nor
are they exceptional. Since the Wardlaw test is not exceptional, it cannot be used as a steppingstone towards the adoption of exceptional solutions to problems of evidential uncertainty as
happened in McGhee.
Having addressed the conceptual aspects of causation in Chapter Two, the remaining chapters
turn towards evidential problems relating to causation. Chapter Three considers the so-called
‘loss of a chance’ argument that was raised by the claimants in Hotson5 and in Gregg.6 These
claimants arose in the context of medical negligence resulting in a delay in the diagnosis and
treatment of patients who were already unwell. Since the patient’s prospects were poor at the
time of the negligence, they faced difficulty in proving that the negligent delay in diagnosis had
caused them to suffer physical harm. Instead they sought to redefine the damage forming the gist
of the action as the loss of a chance of avoiding physical harm. Although this argument centres
on the definition of the relevant damage, its inclusion in this thesis is important for a number of
reasons. The claimant’s attempt to redefine the damage was motivated because of difficulties of
proving a causal link between the negligence and the physical harm, and one of the recurring
issues addressed in this thesis is the relationship between the doctrines of damage and causation.
It is also essential to understand and evaluate the loss of a chance argument because ‘loss of a
chance of avoiding harm’ is often considered to be the equivalent of ‘a material increase in the
risk of harm’ which is contained in the McGhee/Fairchild exception in cases of evidential
uncertainty.
5
6
Hotson v East Berkshire Health Authority [1987] AC 750 (HL).
Gregg v Scott [2005] UKHL 2, [2005] 2 AC 176.
3
In Chapter Four, the discussion turns to an analysis of the approaches taken by courts and
academics to the problem of evidentiary uncertainty in causation, drawing together the themes
running through the previous chapters. It begins by evaluating the evidence relating to the
diseases in McGhee and in Fairchild in order to understand more precisely what the ‘evidentiary
gap’ consists of. The NESS account of causation developed in chapter two enables the
‘evidentiary gap’ to be defined with greater clarity than under the but-for test. The chapter then
engages with the McGhee/Fairchild test of ‘material increase in the risk of harm’ as a test of
sufficient causal connection.
Following McGhee it was suggested that the court had simply taken a ‘robust’ view of the
available evidence, drawing an inference that the negligence had materially contributed to the
harm from the fact that it had materially increased the risk of that harm. It will be argued that the
nature of the evidentiary gap means that it is simply not possible to draw a rational inference of
causation, so this interpretation rests on a fiction.
Following Fairchild, the court in Barker apportioned liability according to the defendant’s
contribution to the total risk to which the claimant was exposed, and Lord Hoffmann
rationalised this on the basis that because the claimant was only able to prove that the defendant
had materially increased the risk of harm, it was the risk of harm rather than the harm itself that
constituted the gist of the negligence action.7 Chapter Four analyses the notion of risk in order to
determine whether risk can, and did, form the gist of the negligence action in that case.
Ultimately it will be argued that if risk exposure was the damage forming the gist of the action,
the damage requirement would be subsumed into the breach inquiry which asks whether the
defendant exposed the claimant to an unreasonable risk of harm. This means that liability would
be based solely on the careless conduct of the defendant so liability would be based on a
retributive form of justice rather than on corrective justice.
7
Barker v Corus (UK) plc [2006] UKHL 20, [2006] 2 AC 572, [35] (Lord Hoffmann).
4
Concluding that a causal link between the negligence and the physical harm cannot be proved in
cases involving an evidentiary gap, the chapter will finally consider the arguments that liability
should nonetheless be imposed because of policy arguments relating to the perceived need to
compensate the victims of mesothelioma. This section will draw on the exposition of the
theoretical foundations of negligence from Chapter One in order to evaluate the policy concerns
and arguments as to the ‘demands of justice’.
Therefore while this thesis takes the analysis of the approaches taken to evidential uncertainty as
its ultimate objective, its contribution to the study of causation extends beyond that narrow
problem. This means that the thesis achieves more than the title might imply, contributing to the
understanding of the basic tests of causation and seeking to add clarity to the relationship
between the causation requirement and the other negligence doctrines, especially the notions of
damage and quantification. It also means that the arguments that are made in relation to the
evidentiary gap are part of an overarching vision of negligence liability and the place of causation
within negligence law, so it makes insights into evidentiary gap that might otherwise be lost, and
gives these arguments a much stronger foundation. Much of the confusion surrounding
causation arises from courts and academics discussing problems of causation too much in
isolation and without giving sufficient weight to the theoretical and doctrinal framework of the
negligence enquiry. This thesis seeks to show that while causation is a complex field, it need not
be confusing.
5
Chapter One: Theoretical and doctrinal framework
Introduction
In order to resolve problematic causal issues in a way that is coherent with the body of
negligence law it is essential to have a clear understanding of the theoretical basis of negligence.
When exceptional approaches to causation are proposed they are usually based on functionalist
arguments such as the need to compensate the claimant or to achieve an economically efficient
solution. It will be argued that these theories cannot provide a coherent basis for negligence so it
is inappropriate to invoke them in difficult cases of causation. The purpose of the first part of
this chapter is, therefore, to set out the legal theory that underlies the tort of negligence. Faced
with a choice between instrumentalist theories and corrective justice-based theories it will be
argued that it is corrective justice that provides the most appropriate theoretical justification for
the tort of negligence. This is because the bipolar structure of the negligence action reflects the
bipolarity of corrective justice and prevents the attainment of instrumentalist conceptions which
are distributive, and therefore multi-polar, in nature. As such the interpersonal morality
encapsulated by corrective justice ought to be reflected in the interpersonal responsibility
enforced through negligence. However, the strict formalism that is characteristic of Weinrib’s
approach to corrective justice is arguably divorced from the practical and realistic need to take
account of how liability decisions impact on wider society.1 It would be inappropriate to insist on
giving effect to interpersonal responsibility through negligence liability where this would have
disproportionately adverse effects on community welfare. As Robertson explains, ‘[i]n the
absence of adequate justification on the basis of [corrective] justice considerations, a duty cannot
be imposed in the public interest, however strong the community welfare justifications’. But, as
1
Peter Cane, ‘Tort Law as Regulation’ (2002) 31 CLWR 305, 310.
6
he adds, ‘a duty can be denied in the public interest, however strongly considerations of
[corrective] justice favour the recognition of a duty’.2
The second source of strain that is placed on the doctrine of causation in exceptional cases is
that the causation requirement is used to address issues that are not really causal and ought to be
addressed by the other negligence doctrines. The chapter will therefore consider how the theory
of corrective justice is translated into concrete legal rules through the various doctrines of the
tort of negligence. It will be seen that factual causation plays a vital but limited role in corrective
justice-based liability, and that it is important to reflect this in the legal rules relating to causation.
Most importantly, non-causal issues relating to the concept of responsibility must be addressed
through the other doctrines. Isolating the role of causation within responsibility in this way will
help to clarify the causal issues in particular cases. Chapter Two will turn to the question of ‘what
causation is’, but first this chapter is concerned with identifying what causation ‘is not’, in other
words with identifying those aspects of responsibility that are not causal and showing how they
are addressed by the other doctrines.
1. The theoretical basis of negligence
It is important to note from the outset that the focus of this thesis is on the tort of negligence so
it is beyond the scope of this work to establish the theoretical foundations of tort law or private
law more generally. What is sought is the theoretical foundation of the tort of negligence.
Englard has contrasted the range of possible approaches:
On the more theoretical level: Purism, formalism, postulates of intelligibility, and
coherence clash with pragmatism and pluralism; on the more ideological, political level:
Individualism, right-oriented personal autonomy contrast with instrumentalism,
utilitarianism, economic efficiency, and social justice.3
Andrew Robertson, ‘On the Function of the Law of Negligence’ (unpublished conference paper, Obligations VI:
Challenging Orthodoxy conference, Canada 18th-20th July 2012)
3 Izhak Englard, Corrective and Distributive Justice: From Aristotle to Modern Times (Oxford University Press 2009) 205.
2
7
Clearly the theories are many and varied. Since this thesis is focused on problems of causation
within negligence, it is beyond the scope of this work to engage in a detailed analysis of all the
possible theories. The objective of this chapter is therefore to signal which of the above values
are most important to negligence and to identify the theoretical position which provides the best
basis for the tort of negligence.
Interpretive legal theories are often evaluated against the four criteria enumerated by Smith: fit,
transparency, coherence, and morality.4 Indeed, it has been said that ‘[a]lthough ‘interpretivism’ is
sometimes treated as a particular camp within private law scholarship, Stephen Smith’s statement
of interpretive legal theory describes what is really an orthodox and widely followed approach to
legal analysis’.5 It should therefore be uncontroversial to refer to these criteria to assess the
theories encountered in this chapter. Robertson has helpfully summarised them as follows:
Those four limbs are: fit (the extent to which the theory is consistent with the outcomes
of cases and possibly also the accepted rules of the given body of law), coherence (the
extent to which the theory reveals an intelligible order in the given body of law and
allows it and related bodies of law to be understood as a unified system), transparency (the
extent to which the theory is consistent with the explanation given by lawmakers, which
in the case of interpreting the common law means the explanations given by judges in the
cases), and morality (‘how the law might be thought to be justified even if it is not
justified’).6
Robertson continues to explain how these criteria are used:
The two most significant points of difference between scholars in the application of the
interpretive method relate to the weight to be attributed to each of the four elements and
the kinds of normative criteria that occupy the place of ‘morality’ in the
framework…Different explanations will score better on different elements. A scholar
Stephen A. Smith, Contract Theory (Clarendon Press 2004) 7-32.
Andrew Robertson, ‘Rights, Pluralism and the Duty of Care’ in Donal Nolan and Andrew Robertson (eds) Rights
and Private Law (Hart Publishing 2011) 438
6 ibid 437-8. Referring to Smith (n4).
4
5
8
must therefore make a choice as to which of the criteria should be weighted more
heavily.7
The reason for preferring corrective justice over pluralist theories is that coherence ought to be
prioritised and, as a monistic theory, corrective justice scores highly for coherence whereas
pluralist approaches are necessarily weak in this respect. Furthermore it will be argued that a
monistic theory based on an instrumentalist goal such as economic efficiency is also lacking in
coherence and moral force. As Wright has explained, coherence is important because without it
the law would be indeterminate:
[W]hen in a particular situation two or more of the pluralistic norms conflict – which
usually will be the case – the theory will be normatively, descriptively and analytically
arbitrary and indeterminate in terms of specifying which competing norm(s) should
predominate, unless there is some foundational norm that can resolve conflicts between
the competing subnorms. Yet if such a foundational norm exists, the theory at its deepest
level is monistic rather than pluralistic.8
Given that coherence is valued to such an extent, it is important to understand more precisely
what coherence entails. Wright continues:
This is not to say that any single norm, no matter how fundamental, can explain and
justify every aspect of the law in general, or any particular area of law…But a successful
normative and descriptive theory of law should at least be able coherently to explain and
justify the principal features of the existing law.9
Beever, whose corrective justice-based approach to negligence prioritises coherence, draws on
Smith’s work to identify two senses in which a theory can be described as coherent: weak
coherence requires the law to be non-contradictory, strong coherence requires that the law ‘can
ibid 438.
Richard Wright, ‘Right, Justice, and Tort Law’ in D. Owen (ed.) Philosophical Foundations of Tort Law (Oxford
University Press 1995) 160.
9 ibid.
7
8
9
be understood as a unified system, perhaps under a single principle’.10 This, he explains, rules out
the notion of ‘limited rationality’ whereby one negligence doctrine such as duty of care could be
based on one principle, and another negligence doctrine such as standard of care could be based
on a contradictory principle. Although each doctrine may display coherence within itself, the
incompatibility of the underlying principles of each doctrine would prevent the tort of negligence
from being coherent in either the strong or weak senses outlined above. This is important to the
scope of this thesis which seeks to understand causation itself but does not treat causation in
isolation from the rest of negligence. Beever further argues that not only should the doctrines be
compatible i.e. display weak coherence, it is desirable that they should be based on a single
unifying principle i.e. strongly coherent:
Because unity leads to greater coherence, and hence to a theory possessing greater
explanatory power, other things being equal, a theory that provides a unified explanation
of the law in the sense elucidated is preferable to one that does not. Unity may not be
mandatory, but it is attractive.11
To summarise the arguments so far, coherence is desirable because of the justificatory force it
carries and because it avoids the indeterminacy inherent in pluralist approaches. It is desirable for
a theory to display ‘strong coherence’ in the sense of enabling the law to be understood as a
unified system, which means that the theory must be able to explain the principal features of the
law even if it is unable to explain every detail of the law.
As mentioned above, Wright correctly asserts that ‘[o]nly a monistic foundational theory holds
out any prospect of being able to do [this]’,12 and he goes on to explain that there are two
principal monistic theories: ‘(1) utilitarian efficiency theory, based on the foundational norm of
Allan Beever, Rediscovering the Law of Negligence (Hart Publishing 2009) 22.
ibid 23-24.
12 Wright, ‘Right, Justice and Tort Law’ (n8) 160.
10
11
10
maximizing aggregate social welfare…and (2) the Kantian-Aristotelian theory of Right or justice,
based on the foundational norm of equal individual freedom’.13
It is the Kantian-Aristotelian theory of justice that provides the more coherent account of
negligence, so the focus of this section of the chapter is to illustrate what the theory consists of
and why it provides the most coherent justification of the tort of negligence. Later sections will
then consider the implementation of this theory through the negligence system and through the
doctrines of negligence law.
1.1
The Aristotelian conception of corrective justice
The starting point to a corrective justice-based account of negligence must be to explain the
Aristotelian conception of corrective justice. It is clear that Aristotle’s account of corrective
justice is abstract and is not an account of the law, let alone an account of a modern tort system,
so it is essential to understand the crucial features of this abstract account that must be translated
into the practices of negligence law in order for the negligence system to be consistent with
corrective justice.
1.1.1
Justice as a whole
Corrective justice, along with distributive justice, was originally elaborated by Aristotle in the
Nicomachean Ethics.14 Aristotle’s discussion of justice follows his exposition of the ‘excellences’ of
character which are intermediate qualities between having too much and too little of contrasting
defects.15 Similarly, in Aristotle’s account, justice must also be an intermediate state. Justice, for
Aristotle, exists as a whole and as a part. Justice as a part is a quantitative state and consists in
‘equal-mindedness’ as opposed to ‘graspingness’ in relation to goods. As Gardner explains,
ibid.
Broadie & Rowe, Aristotle: Nicomachean Ethics: Translation, Introduction and Commentary (Oxford University Press,
Oxford 2002) (cited in this paper as “NE”). For ease of reference, citations will be made to chapter and line
numbers from Aristotle’s original text rather than to pages of this particular translation.
15 NE Books II-IV. For example, courage is an excellence which is the intermediate between fear and boldness.
13
14
11
‘Norms of justice are moral norms of a distinctive type. They are norms for tackling allocative
moral questions, questions about who is to get how much of what’.16 Justice as a whole is a
qualitative concept which comprises justice as a part along with the quality of being a law-abiding
person.17 The law, if it has been laid down correctly, requires a person to display the other
excellences of character (such as courage, moderation, mildness) towards other people. Justice, in
both forms, is an excellence which is a state relative to another individual rather than being an
internal disposition: ‘This justice, then, is complete excellence, only not without qualification but
in relation to another person’.18 Aristotle, having already explored the other excellences of
character in depth, therefore turns to ‘justice as a part’ to address it more fully and it is here that
we are introduced to the subdivision of corrective19 and distributive justice.
1.1.2
Justice as a part: corrective and distributive justice
Justice as a part is a quantitative concept of having one’s share of goods and consists of being
‘equal-minded’:
The unjust person does not always choose the greater share, actually choosing less in the
case of things that are generally bad; but since the lesser evil too is regarded as good in a
way, and what the grasping person wants is more of what is good, this makes people
think of him as grasping.20
It is therefore the intermediate state between having too much and too little both of what is
good and of what is bad. And since justice is ‘other-regarding’ it is relational and consists in
having one’s share compared to another person. Aristotle explains that there are two aspects to
having one’s share:
John Gardner, ‘What is Tort Law For? Part 1: The Place of Corrective Justice’ (2011) 30 Law and Philosophy 1, 6.
NE V.1 (1129a32-34).
18 NE V.1 (1129b26-27).
19 Referred to by Aristotle as ‘rectificatory’ and later by Thomas Aquinas as ‘commutative’.
20 NE V.1 (1129b6-10).
16
17
12
Of the justice that is a part, and of what is just in this sense, one sort is the one found in
distributions of honour, or money, or the other things to be divided up among those
who are members of the political association (for in the case of these things it is possible
for one person to have either an unequal or an equal share in relation to another); while
another is rectificatory, operating in interactions between one person and another.21
The first sort, distributive justice, determines what is just within a community and is an equality
of ‘geometrical’ proportion, while corrective justice determines what is just in an interaction
between two individuals and is an equality of ‘arithmetical’ proportion. Although corrective
justice is the form that corresponds to negligence since it addresses justice in interactions, it can
only be fully understood in relation to distributive justice so both forms will be explained. This is
particularly important since consequentialist justifications of modern negligence law, such as the
utilitarian imperative of maximising wealth which is seen in economic efficiency theories, are
concerned with consequences that affect the community rather than merely the two parties to
the interaction. They are therefore distributive justice-based approaches to negligence so it is
important to understand what distributive justice is, and to illustrate the impossibility of
achieving distributive justice through an institution, such as negligence law, that takes the
structural form of corrective justice.
1.1.2.1 Distributive justice
Distributive justice addresses the distribution of goods among the members of a community. It
is an equality of geometrical proportion, so it does not require every individual to have an equal
share; rather it involves a distribution according to merit. Thus it requires that an individual’s
share of goods (compared to others’ shares) is equal to his level of merit (compared to the merit
of others). Distributive justice does not entail any particular criterion of merit:
21
NE V.2 1130b30-1131a2.
13
[E]verybody agrees that what is just in distributions must accord with some kind of
merit, but everybody is not talking about the same kind of merit: for democrats merit lies
in being born a free person, for oligarchs in wealth or, for some of them, in noble
descent, for aristocrats in excellence.22
To use a common analogy, distributive justice is akin to sharing a cake and determining how big
a slice of cake each person should receive. There is a limited pot of resources since the cake is
finite, and it must be shared according to some criterion of merit, perhaps according to who
contributed the most ingredients or work, who is the most hungry or needy, or all persons
equally. The important point is that for limited resources to be shared in a way that is
distributively just there must be some criterion of merit and a person’s share should be equal to
his worth according to that particular measure of merit. Distributive justice is not, therefore, an
appeal to what the man on the Underground might consider fair.23 It is a comparative exercise,
so is closer in nature to Lord Hoffmann’s concern in White ‘to preserve the general perception of
the law as system of rules which is fair between one citizen and another’,24 although ‘fairness’ is
inevitably still dependent on what criteria of distribution is prioritised.
1.1.2.2 Corrective justice
In contrast, corrective justice concerns the meaning of ‘having one’s share’ in an interaction
between two individuals. Where distributive justice involved equality of geometric proportion,
corrective justice involves an equality of arithmetical proportion.25 This arithmetical proportion is
derived from the pre-transaction equality of the parties:
For it makes no difference whether a decent person has defrauded a worthless one or a
worthless person a decent one…the law pays attention solely to the difference created by
NE V.3 1131a25-30.
Cf White v Chief Constable of South Yorkshire [1999] 2 AC 455 (HL), 495 (Lord Steyn).
24 ibid 511 (emphasis added).
25 NE V.4 1131b25-1132a2.
22
23
14
the damage done, and where one person is committing an injustice, another suffering it,
or one person inflicted damage and another has been damaged, it treats them as equal.26
This notion of equality will be revisited in more depth later, but at this stage it is sufficient to
note that Aristotle regards the parties as equal irrespective of whether they have equal holdings,
irrespective also of whether they are assessed as equal according to any particular criterion of
merit, and irrespective of whether their pre-transaction holdings were distributively just. As
Weinrib has explained, ‘The parties do not have the same quantity of holdings, but they are equal
as the owners of whatever they do have’.27
When one person violates this equality by injuring another he is considered to make a gain which
is equal to the other’s loss, so equality can be restored by requiring him to repair the other’s loss.
It is clear that in concrete terms the gain made by an injurer is not the mirror image of the loss
he inflicts. Aristotle therefore says it may be ‘inappropriate’ to talk of a gain – in layman’s terms
there has been no material gain.28 If I break somebody’s arm I do not ‘gain’ an arm myself. But
because a gain consists not only in having too much of what is good, but also in having too little
of what is bad, I do make a gain through indulging my careless behaviour and through my own
lack of suffering. The gain exists as a violation of the equality of the rights of the parties.29
Because both parties to an interaction are considered equal prior to the interaction, this gain is
correlative to the loss inflicted on the victim. The pre-transaction equality was not an equality of
holdings, nor an equality of status, so a gain is not measured in terms of the concrete increase in
goods or status. The pre-transaction equality was an abstract equality of right to be free from
NE V.4 1132a3-1132a7.
Ernest Weinrib, ‘Corrective Justice’ (1992) 77 Iowa L Rev 403, 408.
28 NE V.4 1132a10-1132a19.
29
Gordley draws on the work of Thomas Aquinas to explain this: ‘Thomas clarified: a “person striking or killing has
more of what is evaluated as good, insofar, that is, as he fulfils his will, and so is seen to receive a sort of gain”. To
gain, then, means to fulfil one’s will. One who has taken or used or harmed another’s resources for his own ends has
‘gained’, and therefore must pay compensation, whether or not his ends have been achieved, and whether or not he
has made a financial gain by pursuing them’ (James Gordley, ‘Tort Law in the Aristotelian Tradition’ in David G
Owen (ed.) Philosophical Foundations of Tort Law (Oxford University Press 1995) 138. Citing Aquinas, In decem libros
ethicorum Aristoteles exposition, (Angeli Pirotta (ed.) 1934), lib. V lectio vi no. 952)).
26
27
15
injury, so the extent of the injury inflicted on another represents the extent to which the
wrongdoer has taken more than he should. As Gordley has explained:
By voluntarily harming the plaintiff, [the defendant] has chosen to use the plaintiff’s
resources for his own ends. The pre-existing equality that corrective justice seeks to
restore is a state in which each party achieves his own goals out of his own resources.30
Again, the concept of equality will be revisited in more detail, but for now it is a sufficient
explanation of the basic Aristotelian account of corrective justice in so far as the key features and
structures of each form of justice have been set out clearly.
In practical terms the correction is made by requiring the injurer to repair the victim’s loss and
this is achieved through compensation because, as Aristotle explains, currency ‘acts like a
measure, making things commensurable’.31
In summary, the key features of corrective justice are that it addresses justice in interactions so it
is bipolar, the two parties are treated as equal before the interaction which gives rise to a
correlative gain and loss and necessitates a correction which is also correlative i.e. the injustice
can only be corrected by requiring the wrongdoer to repair the victim’s loss. As a form of justice
it is distinguished from distributive justice which is multi-polar since it addresses justice across a
community, which achieves a geometrical equality of holdings by specifically recognising the
inequality of individuals according to a criterion of merit, and which necessitates a broad
institution to implement the correct distribution across the community.32
ibid 138.
NE V.5 1133b16
32 It is important to signal these key features because not all ‘corrective justice’ theories of negligence actually reflect
these characteristics. One of the primary problems for Coleman’s ‘annulment’ theory was that it was not based on
correlativity, and this led to him abandoning it in favour of a ‘mixed conception’ of corrective justice which also
incorporates a relational approach. Coleman identified his annulment theory as being one of corrective justice
because, for him, corrective justice required the annulment of wrongful gains and losses. The interpersonal
character of corrective justice was not reflected, however, in Coleman’s account where the victim’s claim for
compensation was based on wrongful loss and the imposition of liability on the defendant was based on wrongful
gain, so the grounds of recovery were conceptually distinct from the grounds of liability. This violated the
correlative character of corrective justice where the grounds of recovery are the same as the grounds of liability i.e.
the loss and gain derive their ‘wrongful’ character from the same source. In his ‘mixed conception’ he incorporates a
relational approach, explaining that the relational conception of ‘wrong’ explains why the wrongdoer and nobody
else has a duty to repair the loss. His work will not be addressed in depth, however, because his approach is still not
based on a single overarching principle as required for strong coherence since he takes a relational approach to
30
31
16
1.2
The moral basis of corrective justice: Aristotelian corrective justice and
Kantian Right
The formalistic reading of the Aristotelian conception of corrective justice has given an insight
into the distinctive nature of corrective justice and the demands of coherence in terms of
maintaining a separation between corrective and distributive justice. But this account is
incomplete insofar as the concept of equality has not been fully explained, so the content and
moral basis of corrective justice remains obscure. It is vital to understand that corrective justice
is not merely a procedural requirement; it is not ‘an empty formalist shell’.33 Beever argues:
According to corrective justice theory, while corrective justice tells us how the injustice
should be corrected, it must also be true that the injustice itself is interpersonal. That is,
the structure of corrective justice must be reflected in the nature of the wrongdoing to
which corrective justice responds.34
This means that corrective justice cannot accommodate an account of wrongdoing that is
distributive, such as aggregate social welfare. Posner has proposed such an approach, arguing
that corrective justice dictates form but not substance so the concept of wrongdoing must be
sought outside Aristotle’s corrective justice and can be based on aggregate social welfare.35 The
formalist reading of corrective justice showed, however, that this would lead to incoherence
because liability would be based on a distributive premise. Strong coherence requires that the
principle of corrective justice is reflected not only in the structure but also in the rules of
negligence liability. Englard has explained, ‘it has to be remembered that even from a strictly
formalist conception, the difference in the formal structures implies some important
explain the wrong, and an annulment approach to explain the loss. Jules Coleman, ‘Corrective Justice and Wrongful
Gain’ (1982) 11 J Legal Stud 421; Jules Coleman, ‘The Mixed Conception of Corrective Justice’ (1992) 77 Iowa L
Rev 427.
33 Wright, ‘Right, Justice and Tort Law’ (n8) 170. Wright also criticises the ‘formalist evisceration’ of corrective
justice: Richard Wright, ‘Substantive Corrective Justice’ (1992) 77 Iowa L Rev 625, 710.
34 Allan Beever, ‘Corrective Justice and Personal Responsibility in Tort Law’ (2008) OJLS 475, 477.
35 Richard Posner, ‘The Concept of Corrective Justice in Recent Theories of Tort Law’ (1981) 10 J Legal Stud 187,
191-201.
17
consequences for the substances. Form and substance become linked through the idea of
coherence’.36
In contrast, Kantian Right, espoused by academics such as Weinrib and Wright, provides an
understanding of wrongdoing that conforms to the requirement of correlativity. As Weinrib
states, ‘the Kantian interpretation fits readily into, and provides content for, corrective justice’.37
Although the ideas of Aristotelian corrective justice and Kantian Right are not explicitly
articulated by courts, ‘they are implicit in [private law] as a coherent justificatory enterprise, in
that they provide its unifying structure and its normative idea’.38 It is Kantian Right that provides
the moral theory, or normative idea, underlying negligence.
In combining corrective justice and Kantian Right, Weinrib begins with Aristotle’s statement that
‘the law treats [the parties to the interaction] as equals however much they may be unequal’ and
discerns three distinct features of corrective justice: first corrective justice treats the parties as
equals; second corrective justice takes no account of any measure of merit of the parties; and
third it focuses on the ‘immediate relationship of doer to sufferer’.39 Weinrib explains that
corrective justice therefore brings together the three ideas of ‘the equality of the parties, the
abstraction from such particulars as social status and moral character, and the conception of
agency present in immediate interaction’.40 The conception of self-determining agency which is
inherent in Kantian Right also displays these three qualities:
The fundamental feature of this agency is the agent’s capacity to abstract from – and thus
not to be determined by – the particular circumstances of his or her situation. Inasmuch
as this capacity is a defining feature of self-determining agency, all self-determining
agents are equal with respect to it.41
Englard (n3) 194.
Weinrib, The Idea of Private Law (Harvard University Press 1995) 19.
38 ibid 19.
39 Weinrib, ‘Corrective Justice’ (n27) 421.
40 ibid 421-22.
41 ibid 422.
36
37
18
This leads him to conclude that Kantian Right ‘dovetails’ with Aristotelian corrective justice.42
So what is Kantian Right, and how does it provide the normative basis for corrective justice?
1.2.1
Kantian Right
Kant’s theory belongs in the tradition of natural right and is based on the idea of free will. Free
will, or freedom, is a characteristic possessed by every rational being and gives every rational
being absolute and equal moral worth:
[M]an regarded as a person, that is, as the subject of a morally practical reason, is exalted
above any price; for as a person (homo noumenon) he is not to be valued merely as a means
to the ends of others or even to his own ends, but as an end in himself, that is, he
possesses a dignity (an absolute inner worth) by which he exacts respect for himself from all
other rational beings in the world. He can measure himself with every other being of this
kind and value himself on a footing of equality with them.43
Respect for the absolute and equal moral worth of everyone gives rise to the ‘categorical
imperative’: ‘act only according to that maxim by which you can, at the same time, will that it
should become a universal law’.44 This is similar to, but wider than, the Golden Rule that you
should treat others as you wish to be treated yourself:
It is morally wrong under the categorical imperative to fail to respect the absolute moral
worth of anyone, including yourself, as a self-legislating rational being, regardless of
whether you would allow others to treat you without proper respect.45
Whereas the Golden Rule allows you to treat others badly if you would be happy to be similarly
treated yourself, the categorical imperative does not allow this because it demands respect for
one’s own moral worth. As Wright explains, ‘[t]he idea of freedom does not imply completely
ibid 422.
Immanuel Kant, The Metaphysics of Morals (Mary Gregor trans., 1991) (1797) cited in Wright (n8) 162.
44 Kant, Foundations of the Metaphysics of Morals (trans. Lewis White Beck) (Bobbs-Merrill Education Publishing 1969)
44.
45 Wright, ‘Right, Justice and Tort Law’ (n8) 162-63.
42
43
19
unrestricted self-determination, but rather self-legislation: self-determination according with
universal law’.46 The exercise of free will in Kant’s theory has both an internal aspect in the
doctrine of virtue, and an external aspect in the doctrine of Right:
The supreme principle of Right is ‘so act externally that the free use of your choice can
coexist with the freedom of everyone in accordance with a universal law’, while the
supreme principle of virtue is ‘[a]ct in accordance with a maxim of ends that it can be a
universal law for everyone to have’.47
So a person who displays virtue desires ends that respect the equal moral worth of all beings and
these ends can therefore be adopted by anyone. ‘Right’ means that even if your intended ends are
not in conformity with respect for the equal moral worth of others you still act in a way that does
respect their equal moral worth.
Therefore Right for Kant, just like justice for Aristotle, is directed towards others. In other
words Right is concerned with the interaction with others and can be enforced through law,
virtue is concerned with the reason for your choice of conduct in this interaction and is thus
internal and cannot be coerced. A virtuous person would act in accordance with universal laws
because he chooses ends that are consistent with those universal laws, but for Right all that
matters is that the person complies regardless of the ends he is pursuing through this
compliance. This seems similar to the distinction between justice as a whole and justice as a part
in Aristotle’s account of justice. The concern of justice as a part is that a person gives others
their due, the motivation behind giving a person his due is not relevant, whereas justice as a
whole is qualitative and requires one to display the other excellences of character towards others.
This concept of equality which is clearly distinct from quantitative assessments also clarifies the
correlativity of the gain and loss in corrective justice:
The relevant loss is the damage suffered by the plaintiff as a result of the defendant’s
failure to give him equal consideration…The defendant’s gain is that he did not carry the
46
47
ibid 162.
ibid 162. Citing Kant (n43) *231, 395.
20
burden of appropriate precautions as he implicated the plaintiff into the web of his own
projects. And the notion of equality against which these gains and losses are measured is
the Kantian prohibition against self-preference in action.48
Kantian Right has important implications for corrective justice because it provides an account of
equality which is divorced from any assessment of the relative merits of the parties according to
a criterion of merit, as well as being independent of the relative holdings of the parties prior to
their interaction.
It therefore illuminates some of the features of corrective justice. In terms of wrongdoing,
Wright says:
Unlike distributive justice, corrective justice does not use interpersonal comparisons or
rankings to implement a relative or proportional equality among the parties to the
interaction…this means that all comparative criteria for determining unjust gains and
losses in corrective justice – including utilitarian, efficiency, and all other aggregative
criteria – are excluded from consideration. This is a powerful substantive implication
from Aristotle’s account.49
This means that the concept of wrongdoing in corrective justice cannot be based on economic
efficiency, because this would be inconsistent with respect for the absolute moral worth of
rational beings since it treats them as means to an end rather than as ends in themselves.
Economic efficiency cannot be accommodated within corrective justice because ‘moral theories
based on equal individual freedom give an absolute and primary status, rather than a contingent
and derivative status, to the autonomy and rights of individuals’.50
So Aristotelian corrective justice and Kantian Right together imply not only the form of the
correction but also the nature of wrongdoing at its core.
Ernest Weinrib, ‘Toward a Moral Theory of Negligence Law’ (1983) 2 Law & Phil 37, 53.
Wright, ‘Substantive Corrective Justice’ (n33) 700-701.
50 Wright, ‘Right, Justice and Tort Law’ (n8) 162.
48
49
21
1.3
Incoherence of consequentialist theories
In contrast to corrective justice, which provides a coherent account of the form of negligence
law and of the wrongdoing at its core, consequentialist accounts are unable to provide a coherent
theoretical foundation. Since they are distributive in nature they cannot be achieved through the
bipolar structure of a negligence action. Moreover, pluralist consequentialist theories, such as the
idea that negligence pursues the twin goals of compensation of victims and deterrence of
wrongdoers, suffer from ‘a surfeit of reasons and norms’, 51 which pose artificial limits on one
another. This means that if negligence doctrines are developed in pursuit of consequentialist
goals the law of negligence will become incoherent. Subsequent criticism of negligence law for
being ineffective or inefficient as a system of compensation or deterrence stems from the fact
that these are distributive goals so are necessarily precluded from being effectively achieved
through the structure of negligence.
Wright explains that compensation and deterrence cannot provide a sound basis for negligence
law because ‘compensation and deterrence of all losses is normatively insupportable,
descriptively implausible, and analytically impossible’.52 He goes on to explain that it is
normatively insupportable because there is ‘no plausible moral argument for requiring others to
compensate every person for every loss no matter how it occurred’.53 It is also impossible
because compensation does not eradicate loss but shifts the loss to others, and because it is
neither possible nor desirable to deter all risky conduct. The bipolar structure of negligence
liability prevents each goal from being fully attained:
The fundamental reason why these criteria are inconsistent with tort law’s correlative
structure is that there is no reason to limit the search for the deepest pocket, or the best
ibid 160.
ibid 159.
53 ibid 159.
51
52
22
loss-spreader, or the cheapest cost-avoider, to the two parties to the tort claim – the
harm-doer and the harm-sufferer.54
Additionally the different distributive criteria impose artificial limits on one another:
When juxtaposed within the tort relationship, compensation and deterrence are mutually
truncating. What limits compensation is not the boundary to which its justificatory
authority entitles it, but the competing presence of deterrence in the same legal
relationship. Thus, tort law compensates victims only when damages serve the purpose
of deterrence. In the same way, tort law artificially restricts deterrence by tying deterrence
not to what is needed to deter wrongdoers, but to what is needed to compensate victims.
In this mixing of justifications, neither goal occupies the entire area to which it applies.
Accordingly, neither in fact functions as a justification.55
Perry objects that compensation and deterrence could be combined in a way that is not mutually
truncating in a system such as compulsory first-party insurance where compensation would be
achieved through insurance payouts and deterrence would be achieved through the setting of
premium levels.56 It is clear, however, that this example is premised on a different institutional
structure, insurance, where a harm-sufferer would be compensated independently of the
insurance premium imposed on the harm-doer. The structure is distributive and matched to the
distributive functions it seeks to achieve. So although Perry may be right to assert that
compensation and deterrence can be combined in a way that is not mutually truncating, it
remains the case that this cannot be achieved through the negligence system which links the
harm-sufferer to the harm-doer in a bipolar relationship.
This means that criticisms targeted at the inefficiency of negligence law as a system of
compensation and/or deterrence ought to be directed at the normative question of whether
there ought to be a system of negligence liability. The significance of this choice is thrown into
Peter Cane, ‘Distributive Justice and Tort Law’ [2001] NZ L Rev 401, 419.
Ernest Weinrib, ‘The Jurisprudence of Legal Formalism’ (1993) 16 Harv J L & Pub Pol’y 583, 587.
56 Stephen R Perry, ‘Professor Weinrib’s Formalism: The Not-So-Empty Sepulchre’ (1993) 16 Harv. J. L. & Pub.
Pol’y 597, 614.
54
55
23
sharp relief by the concerns raised by Atiyah who made an important attack on negligence law as
a system of compensation in The Damages Lottery. The need for financial support clearly extends
beyond those injured by others’ wrongdoing to those suffering due to an accident or natural
causes. As a system for compensation based on need, the negligence system is obviously unjust:
Of all the disabled or handicapped people in society about ten percent suffer from birth
defects, about another ten percent have been injured in accidents, and the remaining
eighty percent are suffering from illnesses and conditions of natural origin. Of the total
number, only about one and a half percent apparently obtain any damages at all. How is
this tiny group selected for preferential treatment?57
But this is a distributive injustice because it is giving preferential treatment to people injured
through someone else’s wrongdoing in a distributive model where the criterion of merit is need.
The assertion is therefore that the negligence system introduces a second, unjustified criterion of
merit according a special status to injury caused by wrongdoing and this has resulted in an unjust
distribution of goods. Arguably Atiyah is criticising negligence unfairly because a system that
takes the form of corrective justice cannot be expected to achieve distributive justice. Atiyah’s
criticism is more serious though, because he argues that the rules of negligence law have been
‘stretched’ by the courts in order to promote compensation.58 In other words, the courts treat
negligence law as a means of achieving compensation so they are adapting the legal rules to try to
achieve compensation in more cases.
As an interpretive theory, Atiyah’s explanation of negligence as a system of compensation is
therefore based more strongly on ‘fit’ and ‘transparency’. As he sees it, the courts reason in terms
of promoting compensation and have shaped the law to compensate more victims of harm. And
the logical conclusion at which he arrives, given that compensation is a distributive model, is that
the negligence system should be abandoned in favour of a fairer compensatory system. It is
important that negligence law adheres to principles of corrective justice to avoid this claim.
57
58
P S Atiyah, The Damages Lottery (Hart Publishing 1997) 143-44.
ibid 32-95.
24
Weinrib warns that ‘a functionalist might regard causation as an indirect way of achieving market
deterrence or some other extrinsic goal, an internal account treats causation as causation, that is,
as a concept that represents the unidirectional sequence from action to effect’.59 Coherence will
be lost if the courts ‘stretch’ the concept of causation in order to promote compensation and
deterrence.
It is important to note that proponents of corrective justice theories of negligence such as
Weinrib and Beever do not generally claim that their theories are normative in the sense that the
law has to have a system of corrective justice-based responsibility. Their theories are interpretive
and therefore compatible with claims that the law of negligence could be abandoned and
replaced with distributive systems of compensation (such as no-fault compensation or insurance)
and of deterrence (such as fines or insurance premiums). Weinrib explains:
To take a modern example, the legal regime of personal injuries can be organized either
correctively or distributively. Correctively, my striking you is a tort committed by me
against you, and my payment to you of damages will restore the equality disturbed by my
wrong. Distributively, the same incident activates a compensation scheme that shifts
resources among members of a pool of contributors and recipients in accordance with a
distributive criterion. From the standpoint of Aristotle’s analysis, nothing about a
personal injury as such consigns it to the domain of a particular form of justice. The
differentiation between the corrective and distributive justice lies not in the different
subject matters to which they apply, but in the differently structured operation that each
performs on a subject matter available to both.60
Their theories do, however, have normative force in that if there is to be a legal system of
negligence liability, in other words a bipolar system where the victim is compensated by the
person whose wrongdoing caused his loss, then the system ought to conform to corrective justice.
Cane explains, ‘it seems clear that Weinrib’s theory is meant to have normative force to the
59
60
Weinrib, The Idea of Private Law (n37) 11-12.
Weinrib, ‘Corrective Justice’ (n27) 415.
25
extent that in his view, if a society chooses to organize some aspect of its life in accordance with
the demands of corrective justice, it should do so consistently’.61 Since the negligence system
takes the form of corrective justice then the content of the legal rules that flesh it out must also
reflect corrective justice, otherwise the system will become incoherent and unjustifiable.
It is therefore clear that Atiyah and Weinrib’s positions can be reconciled since both insist on the
impossibility of achieving distributive justice through a corrective justice structure. The
difference lies in where they proceed from this basis: Weinrib argues that the law should be
matched to the structure, so the rules of negligence law ought to reflect corrective justice; Atiyah
argues that the structure should be matched to the law’s objective, so negligence law ought to be
replaced with a fairer compensation scheme. The problem with Atiyah’s analysis is that it fails to
measure negligence law in its own terms.
It is one thing to acknowledge, as the next section will, that negligence liability has compensatory
and deterrent effects, but it is another thing to say that it has compensatory and deterrent goals
and to measure its success against those goals. As a system of compensation or deterrence it has
been shown that negligence will necessarily be a failure because the goals are mutually truncating
and incompatible with the correlative structure of the negligence action.
1.4
Moving beyond formalism to the legal system - combining Kantian Right
and societal morality.
Corrective justice has been shown to be more than just an ‘empty formalist shell’; it does not
simply proscribe distributive goals but actually prescribes ideas about wrongdoing based on the
Kantian conception of absolute and equal moral worth of rational beings. However, Weinrib’s
formalist conception, in particular his assertion that ‘the purpose of private law is simply to be
private law’,62 has been criticised for being too rigid in its formalism and divorced from the
Peter Cane, ‘Corrective Justice and Correlativity in Private Law’ (1996) 16 OJLS 471, 471-72. Referring to
Weinrib, The Idea of Private Law (n37) 228.
62 Weinrib, The Idea of Private Law (n37) 21.
61
26
realities of the legal system. This is an important and valid criticism because, as already seen,
coherence is not the only quality that is desirable in a justificatory theory.63 The practical
implementation of corrective justice through the negligence system has effects that extend
beyond the parties to society more generally so whilst issues of societal morality cannot be the
driving force behind negligence law they must be taken into account. While it is clear that a
pluralist account cannot provide the premise of negligence law it will be argued that
acknowledging, as Cane does, that the law has distributive effects this does not entail reconceptualising the law as an instrument of distributive justice.
Cane distinguishes between treating consequences as the goals of tort law and acknowledging
them as the effects of tort law:
The fact that private law is not instrumental or distributional in its explicit purposive
orientation does not mean that it does not have instrumental and distributional
effects…To the extent that the traditional view denies that private law has instrumental
and distributive effects, it is wrong. But it would be wrong to think that the common law
leopard has changed (or could - or is likely to – change) its responsibility spots for
instrumentalist and distributional stripes.64
This is an important distinction because it emphasises that while consequentialist theories cannot
provide the premise of negligence law, it might be appropriate to limit the implementation of
corrective justice where the demands of social welfare are particularly convincing. Robertson
adopts this position:
In the absence of adequate justification on the basis of [corrective] justice considerations,
a duty cannot be imposed in the public interest, however strong the community welfare
justifications. But a duty can be denied in the public interest, however strongly
considerations of [corrective] justice favour the recognition of a duty.65
See text to n(4).
Cane, ‘Tort Law as Regulation’ (n1) 330.
65 Robertson, ‘On the Function of the Law of Negligence’ (n2).
63
64
27
The practical implementation of this argument will be fleshed out in section four on the
negligence doctrines; this section makes the preliminary argument that it is theoretically sound to
incorporate social welfare concerns in this limited way.
Weinrib’s account of corrective justice is known for its purism, yet as Morgan has noted it has
been criticised as being ‘so demanding that no actual legal system has ever lived up to it’.66 Cane
explains in more detail that Weinrib’s focus on coherence neglects other aspects of interpretive
legal theory:
An important feature of Weinrib’s ‘formalistic’ approach is that it is concerned with ‘the
law in the books’, and it ignores ‘the law in action’ – in other words, it focuses on tort law
at the expense of what is commonly called ‘the tort system’. As an account of the internal
normative structure of tort law, Weinrib’s idea of correlativity of rights and obligations
(decoupled from Kantian right) seems to me to be essentially correct. On the other hand,
it paints a distorted picture of the way tort law operates in practice.67
This tension can be illuminated by returning to Smith’s four criteria against which to measure
interpretive legal theories: fit, transparency, coherence and morality.68 Weinrib’s account of
corrective justice scores highly in terms of coherence, but it must be remembered that Weinrib’s
account is not entirely an interpretive one. As discussed above, Weinrib’s account is normative
insofar as he argues that if we choose to implement a system of private law then the rules of
private law ought to conform to corrective justice. This means that it does not claim to be perfect
in terms of fit and transparency. But it is important to find a balance between the normative
force of the coherence of corrective justice, and the practical force of the reality of the operation
of the negligence system. This reality is that liability decisions do have redistributive effects
beyond the parties which may generally be considered desirable, but not at the extreme. It will be
Jonathan Morgan, ‘Tort, Insurance and Incoherence’ 67 (2004) MLR 384, 395 referring to arguments of Ken
Kress, ‘Coherence and formalism’ 16 (1993) Harv J L & Pub Pol’y 639.
67 Cane, ‘Tort Law as Regulation’ (n1) 310.
68 See text to (n4).
66
28
argued that the territory on which such a balance can be sought is the remaining criterion in
Smith’s four: morality.
The focus so far has been on an individualistic morality inherent in Kantian Right, the function
of which, Cane says, ‘is to protect and promote the value of human autonomy rather than other
human values such as community and solidarity’.69 Yet, as he argues elsewhere, it would be
morally wrong to ignore the wider social consequences of our actions:
All tort liability is based on some notion of personal responsibility. The fact that matters
of social value are taken into account in tort law does not alter this; it is simply an
outworking of the straightforward moral principle that a man’s responsibilities are partly
a function of the fact that man is a social being.70
So whilst coherence should be prioritised, this should not be at the expense of morality.
Although Weinrib contends that distributive and corrective justice cannot be integrated into ‘any
overarching form’,71 it will be argued that it is wrong to ignore societal morality completely,
especially once we acknowledge that the practical operation of the negligence system does have
effects that extend beyond the parties to the case. Whilst the focus on autonomy is consistent
with the concept of Kantian Right which lies at the heart of corrective justice and ought rightly
be prioritised, it is too extreme and fails to reflect man’s regard for society. At some point
respect for the moral worth of the individual crosses a line to become so extreme that it
disproportionately impacts on the collective interests of the other members of a society.
It could be argued that Cane is simply observing that man interacts on a societal level as well as
on an individual level and that this is the equivalent of Aristotle’s explanation that there are two
forms of justice to correspond to these two forms of interaction. But Cane is making the
important point that a person does not draw a clear, distinct line between his interactions with
individuals and with the community, or rather, a person does not make an absolute distinction
Peter Cane, ‘Rights in Private Law’ in Donal Nolan and Andrew Robertson (eds) Rights and Private Law (Hart
Publishing 2011) 56.
70 Peter Cane, ‘Justice and Justifications for Tort Liability’ (1982) 2 OJLS 30, 62.
71 Weinrib, ‘The Jurisprudence of Legal Formalism’ (n55) 589.
69
29
between his reasons for acting in the two different contexts. Beever identifies four different
‘spheres of morality’: personal, interpersonal, societal and international.
Personal morality asks how an individual should behave in order to be a good person.
Interpersonal morality considers interactions between one individual and another. The
focus here is on how persons should conduct themselves vis-à-vis one another as two
individuals rather than as isolated individuals or as members of a collective. The third
sphere is societal. The concern here is how to govern society and how to regulate the
behaviour of individuals for the common good. The final sphere is the international. This
sphere considers the impact of our actions on everyone, whether they belong to our
political community or not, or the impact of our state on other states.72
For Beever, negligence liability is based on corrective justice so it reflects interpersonal morality.
As such, interpersonal morality is reflected in key features of negligence law, for example the
objective standard of care arises because negligence is a form of interpersonal morality rather
than personal so it does not matter in negligence whether there was subjective
blameworthiness.73 Thinking of morality as operating in different spheres can therefore help to
understand the different concerns relevant to each. But it would seem strange to think that these
spheres of morality actually operate in isolation from one another. Personal morality informs
interpersonal morality; its equivalent in the Kantian approach is virtue which informs Right
through the shared concept of conformity with universal law. There should be a similar exchange
of ideas between societal and interpersonal morality.
This is an important limit to the coherence of corrective justice when it is implemented through
the practices of negligence law, and it risks appearing indeterminate and therefore blurring the
line between showing an awareness for avoiding extreme redistributive effects of liability and
actually adopting redistribution as a goal of liability. Yet the reality is that a theory of negligence
is measured not only in terms of coherence but also in terms of morality, fit and transparency.
72
73
Beever, Rediscovering (n10) 42.
Beever, ‘Corrective justice and personal responsibility’ (n34) 491.
30
While fit and transparency may be descriptive, both coherence and morality have prescriptive
force and the effort should be made to combine them. The account of negligence set out in this
thesis is still based on corrective justice but takes a view of interpersonal responsibility that
combines autonomous and societal moral values. It is a moral account that is premised on the
absolute and equal moral worth of beings, but which shows an awareness that individual liability
decisions can sometimes effect an extreme change to the distributive pattern. Whilst
consequentialist accounts are distributive and thus cannot provide a coherent basis for
negligence we cannot ignore the effects that liability has and should remember that man, as a
societal being, will draw the line somewhere and say that his individual needs should not be
prioritised over the needs of others when the effects become too significant. So while social
welfare considerations cannot ground liability, exceptionally they may limit it.74
A final clarification is needed in relation to terminology. It is important to refer to this corrective
justice-based approach as reflecting ‘interpersonal responsibility’. Morgan has referred to the
principles of ‘individual responsibility and corrective justice’ as distinct from collective
responsibility, and Cane has said that ‘tort law is a set of rules and principles of personal
responsibility’.75 Beever equates personal responsibility with personal morality which, as
explained above, is distinct from interpersonal morality and responsibility. He explains that
personal responsibility would be undermined by liability insurance because it is not the defendant
personally who compensates the claimant, whereas corrective justice requires only that the
defendant see to it that the claimant is compensated and ‘does not imply any moral claim about
the defendant per se.76Although ‘interpersonal responsibility’ is a more cumbersome expression
than ‘personal responsibility’ it should be used to avoid ambiguity when referring to corrective
justice-based liability.
Robertson ‘On the Function of the Law of Negligence’ (n2).
Cane, ‘Distributive Justice’ (n54) 403.
76 Beever, ‘Corrective Justice and Personal Responsibility’ (n34) 495.
74
75
31
1.5
The place of causation in corrective justice-based interpersonal
responsibility
Causation has been largely unmentioned up until this point because the focus has been on the
theoretical foundations of negligence generally rather than on the theory of causation. The task
now is to illustrate that causation is a central feature of corrective justice-based interpersonal
responsibility and that this centrality is unaffected by the argument that the distributive effects of
liability ought to be taken into account in order to balance coherence and morality. As a
conception of responsibility, however, causal responsibility is not synonymous with interpersonal
responsibility, so causation occupies a vital but limited role which must be supplemented by
ideas about wrongdoing and wrongful loss.
1.5.1
The pivotal place of causation in corrective justice
Causation is central to corrective justice because corrective justice concerns justice in interactions
and without causation there is no transaction between the parties to the interaction as Weinrib
explains:
The requirement of factual causation establishes the indispensable nexus between the
parties by relating their rights to a transaction in which one has directly impinged upon
the other. Tort law does not typically pursue wrongful conduct in the abstract. It
concerns itself with such conduct only when it materializes in harm to a given person so
that compensation can flow from a particular tortfeasor to his particular victim. 77
The causation requirement in negligence is therefore crucial. It is clear from the speeches in the
House of Lords’ decision in Chester v Afshar, 78 that if causation is not established liability is no
longer based on corrective justice. The defendant doctor in that case breached the duty of care
owed to his patient by failing to warn her of a small risk inherent in the proposed operation. This
was an unavoidable risk that would have existed even if the operation itself was carried out with
77
78
Weinrib, ‘Toward a Moral Theory (n48) 38.
Chester v Afshar [2004] UKHL 41, [2005] 1 AC 134.
32
appropriate care. Unaware of the risk, the claimant underwent the operation which was carried
out without negligence but the risk materialised resulting in injury. In negligence it must be
established that the defendant’s breach of duty, in this case his failure to warn of the risk, was a
cause of the claimant’s injury. The difficulty in this case was that the claimant very honestly
admitted that even if she had been warned of the risk she may still have undergone the operation
after seeking a second opinion. In other words, the defendant’s failure to warn her of the risk
made no difference to her decision to undergo the operation so it arguably made no difference
to the outcome. Although it has been suggested that a causal link can be established,79 a
substantial body of academic opinion maintains that this was a case where the defendant’s
negligence was not a cause of the claimant’s injury.80 Indeed the speeches in the House of Lords
generally proceed on the basis that causation could not be established.81 The case is interesting at
this stage because the suggestion that liability was imposed in the absence of a causal link means
that it raises issues regarding the place of causation within interpersonal responsibility.
Lord Bingham found that the claimant had failed to satisfy the but-for test of causation and
asked ‘whether Miss Chester should be entitled to recover even though she cannot show that the
negligence proved against Mr Afshar was, in any ordinary sense, a cause of her loss’.82 Ultimately
he and Lord Hoffmann adhered to traditional corrective justice-based negligence principles and
concluded that the defendant should not be held liable in the absence of a causal link.
Given the apparent lack of a causal link between the doctor’s failure to warn and the patient’s
injury, the reasons given in the remaining judgments in favour of liability are premised on ideas
Stapleton has argued that the causal link exists because if the patient had been warned of the risk she would have
sought a second opinion so the operation would have taken place on a different day. The risk itself remains constant
at one to two percent, so on the balance of probabilities the risk would not have materialised on the alternative
occasion: Jane Stapleton, ‘Occam’s Razor Reveals an Orthodox Basis for Chester v Afshar’ (2006) 122 LQR 426.
80 See for example Sarah Green, ‘Coherence of Medical Negligence Cases: A Game of Doctors and Purses’ (2006)
14 Med LR 1, 4.
81 It could be argued that Chester is a case where the harm should be reconceptualised in order to avoid the causal
problem. This idea which will be developed in later chapters where it will be argued that in very exceptional cases
considerations of interpersonal responsibility require the reconceptualization of the concept of damage to include
the loss of a chance. It will ultimately be argued, however, that Chester is an unexceptional case so there is no
justification for regarding the claimant as having lost a chance, or for regarding the interference with her right to
give informed consent as damage in itself.
82 Chester (n78) [9] (Lord Bingham).
79
33
that are divorced from corrective justice. Lord Steyn focused on vindicating the right of the
patient to give informed consent, explaining that ‘not all rights are equally important. But a
patient’s right to an appropriate warning from a surgeon when faced with surgery ought
normatively to be regarded as an important right which must be given effective protection
wherever possible’.83 The purpose of this right is not primarily to protect the welfare of the
patient, but to ensure that ‘due respect is given to the autonomy and dignity of each patient’. 84
Yet he does not note that vindication of this right would require liability whenever a doctor fails
to warn a patient of the risk regardless of whether the risk materialises in harm. The damage
requirement in negligence imposes an artificial limit on this vindicatory imperative.
Lord Steyn also considered it important that it was the doctor’s act of operating, if not his
negligence in particular, that caused the harm, but Lord Hoffmann explained that this amounts
to treating the doctor as an insurer.85 He also explained that ‘it would excuse the doctor in a case
in which he had a duty to warn but the actual operation was perfectly properly performed by
someone else, for example, by his registrar’.86 If this occurred, the compensatory objective would
be truncated by the bipolar structure which prevents the court from imposing liability on the
‘someone else’. Causation is clearly pivotal to corrective justice; there cannot be corrective
justice-based liability in the absence of a causal link but, as Weinrib’s formalism demonstrated,
distributive arguments for liability cannot be achieved within a bipolar, corrective justice-based
institution.
Lord Hope objected that:
A duty was owed, the duty was breached and an injury was suffered that lay within the
scope of the duty. Yet the patient to whom the duty was owed is left without a remedy.87
ibid [17] (Lord Steyn).
ibid [18] (Lord Steyn).
85 ibid [35] (Lord Hoffmann).
86 ibid [35] (Lord Hoffmann).
87 ibid [73] (Lord Hope).
83
84
34
Yet this is the nature of corrective justice. Corrective justice is not concerned with punishing
wrongdoing and it is not primarily concerned with compensating loss. Corrective justice is
concerned with wrongdoing that results in loss. The causation requirement has been criticised as
introducing an element of ‘moral luck’ into negligence liability.88 Howarth suggests it is ‘a kind of
privilege for defendants, a privilege that allows defendants to escape liability because of
coincidences for which they can take no credit’.89 If the focus of liability was the wrongdoing of
the defendant alone then there may be some force to this argument. But the focus of corrective
justice is not on either of the parties alone, it is on the inequality that occurs when the
wrongdoing of one person causes a loss to another person, in other words it is only concerned
with equality in transactions and when two people interact there is only a transaction when
something passes between them i.e. one causes a loss to the other.
As explained above, corrective justice theory is normative insofar as once the law adopts a
system, such as negligence law, that reflects the bipolar structure of corrective justice then the
legal rules ought to conform to corrective justice otherwise they become unjustified and
indefensible. This discussion of the causation requirement in Chester v Afshar has shown that in
the absence of a causal link the courts adopt justifications for liability that are based on reasons
of distributive justice which are therefore incapable of providing a coherent explanation of the
basis of liability, and the effects of which are artificially limited by the corrective justice structure
of negligence liability. Causation is clearly indispensable to the operation of corrective justice.
1.5.2
The limited role of causation
As noted above, Lord Steyn considered it relevant in Chester that it was the defendant’s act of
operating, if not his negligence in particular, that caused the claimant’s injury. Honoré has
suggested that in these circumstances the doctor is morally responsible for the outcome of his
David Howarth, ‘Many duties of care – or a duty of care? Notes from the underground’ (2006) 26 OJLS 449, 461.
See also Jeremy Waldron, ‘Moments of Carelessness and Massive Loss’ in David G Owen (ed.) Philosophical
Foundations of Tort Law (OUP 1995) 388.
89 Howarth (n88) 461.
88
35
non-negligent act.90 He calls this ‘outcome responsibility’.91 He argues that people of full capacity
are morally responsible for the outcomes of their voluntary actions even if the outcomes are
unintended, unforeseen or even unforeseeable. Outcome responsibility is not, therefore, an
explanation of negligence liability and it allows people to take both credit for the positive
outcomes and discredit for the negative outcomes of their actions. Instead, by showing that
people are morally responsible for the outcomes of their actions, Honoré seeks to demonstrate
that legal liability that is not based on moral blame (i.e. strict liability and the application of the
objective standard of care to those unable to achieve that standard) is still a form of moral
responsibility.
Honoré argues that outcome responsibility is ‘crucial to our identity as persons’, without it ‘we
could have no continuing history or character’.92 But as Cane argues, if outcome responsibility
was equated with causal responsibility this would be equally damaging to our personality/agency:
If people could not claim, and did not have to accept, ownership of any conduct or
outcome which was affected by luck, we would lose a secure sense of ourselves and
others as agents whose conduct can have effects in the world; and we would end up
seeing ourselves as ‘victims of circumstance’. On the other hand, entirely ignoring the
role of luck in our lives would be equally damaging to our sense of personal identity. If
people could take credit for all good outcomes of their conduct no matter how lucky,
and were held responsible for all bad outcomes of their conduct no matter how unlucky,
we would lose the feeling of being able to act purposively and to influence
events…Consequently, we can claim ownership of some of the lucky outcomes of our
Tony Honoré, ‘Medical non-disclosure: causation and risk: Chappel v Hart’ (1999) 7 Torts L J 1.
See generally Tony Honoré, ‘Responsibility and Luck’ in Tony Honoré, Responsibility and Fault (Hart Publishing
1999).
92 ibid 29.
90
91
36
conduct, but not all; and we must accept ownership of some of the unlucky outcomes of
our conduct, but need not accept all of them.93
The problem with outcome responsibility is therefore that it contains an unarticulated limit on
the consequences that will be considered to be outcomes of a person’s conduct. Honoré’s
outcome responsibility is based on his particular approach to causation ‘according to which our
interventions in the world are cut short by later interventions, so that each person has a set of
achievements or failures that is exclusively their own or is shared with a few others’.94 He
considers that ‘[h]ow far one should go back in a causal inquiry depends on the purpose of the
inquiry’.95 This introduces an element of circularity to Honoré’s account as Cane highlights:
For the purposes of attributing responsibility, then, principles of causation are
inextricably linked with principles of responsibility. And just as what counts as a cause
depends on the purpose of the causal inquiry, so the answer to the question of what we
are responsible for will depend on the purpose of the inquiry…In short, the scope of
outcome responsibility…depends on the purposes it serves.96
It is clearly not possible for outcome responsibility to determine objectively which outcomes a
person is responsible for. Instead, outcome responsibility requires a value-judgment about which
outcomes a person ought to be considered responsible for, and it is this normative aspect that
needs to be justified. So to say that the doctor in Chester was morally responsible for the harm
caused by his carefully executed operation on the basis of outcome responsibility is merely to say
that his action was a cause of the harm and the purpose of the vindicating the patient’s right to
know/enforcing the doctor’s duty to warn through the negligence inquiry justifies the attribution
of responsibility. Effectively this amounts to saying that if the purpose of the inquiry is
vindication/deterrence rather than corrective justice then there is a reason for concluding that
Peter Cane, ‘Responsibility and Fault: A Relational and Functional Approach to Responsibility’ in Peter Cane and
John Gardner (eds.) Relating to Responsibility, Essays in Honour of Tony Honoré on his 80 th Birthday (Hart Publishing 2001)
91.
94 Tony Honoré, Responsibility and Fault (Hart Publishing 1999) 10.
95 ibid 11.
96 Cane, ‘Responsibility and Fault’ (n93) 90.
93
37
the doctor is morally responsible for the outcome. It is not the force of his moral responsibility
for the outcome that justifies the imposition of liability.97
In corrective justice a person is not responsible for all loss that they cause, only for wrongful
loss. Causation thus provides the essential link between the parties so it is the starting point for
being able to find moral responsibility but its role is limited because it cannot explain where
moral responsibility for outcomes ends. It is the concept of wrongdoing that provides the
dividing line between those losses that are wrongful and those that are not, and this must be
addressed through the remaining negligence doctrines other than causation.
2. The doctrinal framework
Having established that negligence should be seen as a corrective justice-based system, i.e. a
system premised on interpersonal responsibility and a conception of morality that takes seriously
the intrinsic moral worth of people, and having seen the place of causation within interpersonal
responsibility, it is important to turn to the question of how this is translated into practice
through the doctrines of negligence law. As explained in the introduction to this chapter, many
problems that arise in exceptional approaches to causation exist because the doctrine of
causation is used to address other aspects of responsibility which are not truly causal. It is
essential to clarify ‘what causation is not’ and to illustrate clearly that evaluative aspects of
responsibility are addressed through the other negligence doctrines.
Coherence ‘signifies a mode of intelligibility that is internal to the relationship between the parts
of an integrated whole’,98 such that ‘[t]he negligence concepts form an ensemble that brackets
and articulates a single normative sequence’.99 Having shown that causation has a vital yet limited
role within the concept of interpersonal responsibility it is essential that this is reflected in the
negligence doctrines. The overall objective of this section is therefore to show the
The relationship between causation and responsibility will be revisited in chapter two on causation.
Weinrib, The Idea of Private Law (n59) 14.
99 Weinrib, ‘Legal Formalism’ (n55) 593.
97
98
38
interrelationship between causation and the other negligence doctrines in order to emphasise the
role played by the other doctrines in determining liability together.
2.1
The duty of care
The duty of care dominates this section because it is there that the issues of interpersonal
responsibility and wider distributive arguments come to the fore. It was argued in section 1.4 that
it is theoretically sound to maintain that liability should only be imposed if it is grounded in
considerations of interpersonal responsibility although it may exceptionally be limited to avoid
distributive injustice. This section therefore proceeds to flesh out this argument by evaluating the
concrete issues that motivate duty decisions.
2.1.1
Duty as the ‘central organising concept’100
Coherence requires that the negligence doctrines come together to form an integrated whole. A
virtue of Beever’s corrective justice-based approach to negligence is that it seeks to achieve this:
[T]he ‘principled approach’…treats the stages of the negligence enquiry as forming a
unified investigation. This investigation is an enquiry into whether the unreasonable risk
created by the defendant (standard of care) was an unreasonable risk of injury to the
claimant (the duty of care) and an unreasonable risk of the injury that the claimant
suffered (remoteness).101
Treating the negligence doctrines as forming a ‘unified investigation’ reflects Weinrib’s concern,
noted above, that the negligence doctrines ‘articulate a single normative sequence’.102 However,
Beever addresses the standard of care before he addresses the duty of care. The danger of this is
that the duty doctrine risks appearing as a control mechanism rather than as a meaningful
concept and this does not sit comfortably with a moral account of negligence liability.
This label is used by Nicholas J McBride, ‘Duties of care – do they really exist?’ (2004) 24 OJLS 417, 423-24.
Beever, Rediscovering (n10) 115.
102 Weinrib, ‘Legal Formalism’ (n55) 593.
100
101
39
If the duty doctrine is merely a control mechanism, or is merely a duty to compensate a victim if
one’s carelessness results in loss, it would lose its moral force. McBride explains that in this kind
of ‘cynical’ view of duty, ‘[t]he law of negligence would no longer purport to guide people’s
behaviour…Instead, the law of negligence would simply attach costs to certain forms of
behaviour and leave it up to individual citizens to decide whether or not to participate in those
forms of behaviour’.103 This cynical view is incompatible with Kantian Right since it treats the
claimant (and potential claimants) merely as a commodity or as a cost to be entered into an
equation. Corrective justice is a moral theory so it implies an idealist approach to duty as a
meaningful duty to take care since this takes seriously the moral worth of people.
McBride goes on to argue that the duty is the ‘central organizing concept around which the
whole of the law of negligence revolves’ in an idealist account:
After all, an idealist would say, a claimant will only be entitled to sue a defendant in
negligence if: (1) the defendant owed her a duty of care; (2) the defendant breached that
duty of care; (3) the breach of that duty of care caused her to suffer some kind of loss; and (4)
the loss suffered by the claimant as a result of the defendant’s breach was the kind of loss
which the duty of care breached by the defendant was imposed on him in order to avoid.104
Logically, therefore, duty must be analysed before the other doctrines. This is important because
the arguments raised at the duty stage focus on the relationship between the parties, so they
bring to the foreground the considerations of interpersonal responsibility in that particular
relationship. Where problems arise in the remaining doctrines, and a solution is sought that
forms part of a coherent approach to negligence, then we must go back to the issues of
interpersonal responsibility identified at the duty stage and carry them through into the solution
to the problem.
McBride illustrates the centrality of the duty of care in establishing the interpersonal aspect of
responsibility and its role in informing the remaining negligence doctrines through cases such as
103
104
McBride (n100) 424.
McBride (n100) 423-24.
40
Reeves v Commissioner of Police of the Metropolis.105 This case involved the duty of care owed by the
police to a prisoner in their custody to take reasonable care to prevent him committing suicide.
The police failed to take reasonable care and the prisoner took his own life. Ordinarily the
voluntary and unreasonable act of the victim in harming himself would be considered at the legal
causation stage of the negligence inquiry to have broken the chain of causation between the
defendant’s negligence and the victim’s harm. But in the context of this particular duty of care, a
duty specifically targeted towards preventing the victim from self-harm, the victim’s voluntary
and unreasonable act of self-harm cannot be considered to break the chain of causation. This is
not merely a matter of reinforcing the duty owed by the police in the way that imposing liability
on the doctor in Chester v Afshar in the absence of a causal link was directed towards enforcing
the content of the doctor’s duty to warn. In Reeves there was a causal link, and the issues of
interpersonal responsibility showed that the coherent solution was to hold the harm to fall within
the scope of this responsibility. The duty of care similarly informs the remaining negligence
doctrines in cases where harm is inflicted by a third party.106
Davies also draws on the idea that the duty of care and the remaining negligence doctrines form
unified investigation in order to explain the evolution of the test for remoteness.107 He suggests
that the approach adopted in Re Polemis and Furness Withy Co,108 asking whether the harm was a
direct consequence of the negligence regardless of whether it was reasonably foreseeable, was
acceptable given the narrow approach courts took to duty of care at that time. Since Re Polemis
was decided before Donoghue v Stevenson,109 the duty relationship had to be shown to fall within an
established duty situation, what Davies refers to as a ‘particularised, relational’ duty of care,110 so
Reeves v Commissioner of Police of the Metropolis [2000] 1 AC 360 (HL).
Claire McIvor, Third Party Liability in Tort (Hart Publishing 2006) 18-19.
107 Martin Davies, ‘The Road from Morocco: Polemis through Donoghue to no-fault’ (1982) 45 MLR 534.
108 [1921] 3 KB 560.
109 [1932] AC 562 (HL(Sc)).
110 ibid 541. In this instance, Davies says, the duty was based on the authority of Dean v Keate (1811) 3 Camp. 4
(170 ER 1286) which established that a bailee owes the bailor of property a duty to exercise reasonable care over the
property he has hired.
105
106
41
‘a wide remoteness test was unexceptionable when the duty was narrowly conceived’.111 This
remoteness test of ‘directness’ was later overruled in The Wagon Mound112 in favour of a test of
reasonable foreseeability of the type of harm. This shift, Davies suggests is explained in part by
the expansion of the duty of care in Donoghue v Stevenson. The more restrictive nature of the Wagon
Mound test was better suited to limiting liability given the less restrictive stance to establishing
duty of care which was then being taken applying Lord Atkin’s ‘neighbour principle’. Given the
more restrictive incremental approach that is now being taken towards duty of care under the
Caparo test,113 and the return to what Stapleton labels ‘pockets of liability’,114 this may explain why
courts are beginning to look towards the ‘scope of duty’ as the approach to remoteness.115 Such a
view is supported by Steele who has suggested ‘[i]n more recent years, the concept of ‘duty’ has
become more specific to the particular circumstances. Perhaps it is time to update The Wagon
Mound, to reflect this change in the approach to duty’.116
Having established that the duty of care is central to shaping the rest of the negligence inquiry
because of the role it plays in identifying the particulars of the relationship of interpersonal
responsibility between the parties to the case, it is important to turn to the duty itself and
consider which duty factors are truly reflective of interpersonal responsibility.
2.1.2
Identifying issues of interpersonal responsibility in the duty doctrine
This section examines the duty doctrine to identify those duty factors that reflect interpersonal
responsibility and ought to influence the approach to liability questions including
causation/damage, and to filter out those that reflect social welfare/distributive justice and are
therefore incapable of grounding liability. It was argued in the theory section that this approach
Davies (n107) 541. He also argues that the social context of Re Polemis made the decision desirable at the time.
Overseas Tankship (UK) Ltd v Morts Dock and Engineering Co Ltd (The Wagon Mound (No1)) [1961] AC 388 (PC)
113 Caparo Industries v Dickman [1990] 2 AC 605 (HL).
114 Jane Stapleton, ‘Duty of care and economic loss: a wider agenda’ (1991) 107 LQR 249.
115 South Australia Asset Management Corp (SAAMCo) v York Montague Ltd [1997] AC 191 (HL), 210-13 (Lord
Hoffmann).
116 Jenny Steele, Tort Law: Texts, Cases and Materials (2nd ed., Oxford University Press 2011) 211.
111
112
42
is appropriate, so the current section seeks to flesh out the various considerations to show what
it means in concrete terms. This is because it is the issues of interpersonal responsibility that will
be prioritised when considering exceptional approaches.
The dividing line between considerations of interpersonal responsibility and distributive factors
is sometimes said to be the line between ‘principle’ and ‘policy’, or the distinction between
‘foreseeability and proximity’ and ‘fairness, justice and reasonableness’. Since these terms do not
have any agreed meaning it is necessary to give concrete examples of a range of considerations
that do/do not reflect interpersonal responsibility.
Beever distinguishes between ‘principle’ and ‘policy’ although this division has been criticised by
legal realists, notably Stapleton, who argue that there is no distinction to be drawn and that
reasons for liability will be labelled ‘principle’ to make them seem palatable and ‘policy’ to make
them seem political. Stapleton argues that ‘appeal merely to so-called ‘principles’ often simply
masks the substance of a judge’s reasoning process’.117 Witting has similarly criticised Beever’s
corrective justice-based account of negligence on the basis that ‘Beever provides very little
comfort to those who prefer principle when ‘principle’ is little more than Beever says it is –
namely that law as determined by Beever’s own application of reflective equilibrium’. 118 However
Beever’s exposition of ‘principle’ is not nearly as subjective as Witting suggests. Beever explains
that:
‘Principle’ refers to the rules and doctrines of the law itself… ‘Policy’, on the other hand,
can be defined only negatively. ‘Policy’ is everything apart from principle. For example,
policy has been held to include issues of distributive justice, social morality, economic
efficiency, public opinion and so on. But it is impossible to define the content of these
terms exactly, because people disagree on what constitute the rules and doctrines of the
Jane Stapleton, ‘The Golden Thread as the Heart of Tort Law: Protection of the Vulnerable’ (2003) 24 Aust Bar
Rev 135, 136.
118 Christian Witting, ‘The House that Dr Beever Built: Corrective Justice, Principle and the Law of Negligence’
(2008) 71 MLR 621, 630.
117
43
law. Accordingly, there is no ‘theory neutral’ way of defining the content of principle and
policy.119
In other words, the distinction between principle and policy is dependent on the foundational
theory of negligence that one prefers. So ‘principle’ is not simply ‘what Beever says it is’;
‘principle’ is the set of rules that express corrective justice. To an extent this is subjective because
it depends on Beever’s account of what corrective justice is. But it does not suffer from the
indeterminacy and subjectivity that Witting implied. Indeed, although Witting does not
necessarily draw such a clear line between principle and policy, he has argued elsewhere that
there is a meaningful distinction to be drawn between the three elements of the Caparo test for
duty of care, notably between ‘proximity’ and consideration of what is ‘fair, just, and reasonable’.
He suggests that proximity looks inwards to the relationship between the parties and ‘fair, just
and reasonable’ looks outwards to ‘the potential impact that the recognition of a duty might have
upon a number of parties who are not before the court’.120 This is broadly correct but the dividing
line is not quite so clear and bright.
Robertson’s account is particularly useful, especially since he advocates the broad approach
adopted in this thesis. Robertson prioritises justice between the parties but, in contrast to Beever
whose pure corrective justice account is ‘designed to eliminate both the need for appeal to policy
considerations and the permissibility of that appeal’,121 Robertson ‘recognises that community
welfare considerations can come into play in a secondary way to negate the duty of care in
certain limited circumstances’.122 He draws out considerations across all three stages of the Caparo
test that address ‘justice between the parties’ and considerations at each stage that concern
‘community welfare’.123 In other words, he draws a broad distinction between ‘justice between
Beever, Rediscovering (n10) 3.
Christian Witting, ‘Duty of care: an analytical approach’ (2005) 25 OJLS 33, 38.
121 Beever, Rediscovering (n10) 29.
122 Roberston, ‘Rights, Pluralism and the Duty of Care’ (n5) 441.
123 ibid 441-4.
119
120
44
the parties’ which is the focus of corrective justice-based interpersonal responsibility, and
‘community welfare’ which is distributive.
In terms of justice between the parties, Robertson says that there are three ‘stages’: reasonable
foreseeability, proximity, and ‘other considerations of justice between the parties’. Reasonable
foreseeability is widely regarded as a necessary element since it signals that the defendant at least
had capacity to act differently – it makes the first step away from purely causal responsibility
towards a form of moral responsibility for outcomes. ‘Proximity’ is not susceptible to objective
definition, but as Robertson argues ‘even if the concept of proximity does no more than to give
focus to a particular aspect of the duty inquiry and to express the nature of what is in issue in the
assessment of a discrete and distinctive set of considerations, that may be considered enough to
justify its retention’.124 Using similar terms to Lord Atkin’s formulation in Donoghue v Stevenson, he
explains that proximity ‘is essentially concerned with the question whether the claimant was so
closely and directly affected by the defendant’s conduct that the defendant ought to have had the
claimant in mind when considering whether and how to act’.125 Proximity factors include:
the extent of the defendant’s control over the risk, whether the defendant created the
risk, the extent of the defendant’s knowledge of the risk, the magnitude of the risk and
degree of foreseeability of the harm, the extent of the claimant’s vulnerability to harm
arising from the risk, the extent to which the claimant relied on the defendant, whether
the defendant knew of that reliance and whether the defendant can be said to have
assumed a responsibility, either to the claimant or for a particular task.126
Robertson also notes that proximity allows the court to look at the scope and limits of the
claimant’s personal responsibility and the defendant’s autonomy. The availability of insurance is
irrelevant here because as Robertson explains it is an alternative compensatory remedy which
does not actually render the claimant less vulnerable to harm, just more able to deal with the
Andrew Robertson, ‘Justice, community welfare and the duty of care’ (2011) 127 LQR 370, 374.
ibid 375.
126 ibid.
124
125
45
financial consequences of harm. The ‘other considerations of justice between the parties’ that
Robertson identifies are first whether a duty in negligence would be inconsistent with the
contractual context between the parties, and second whether recognition of a duty would impose
an unreasonable burden on the defendant. This second consideration seems most controversial,
because as Stapleton has suggested, the fact that a defendant might be exposed to a large (as
opposed to indeterminate) number of claims is an unconvincing reason to deny liability.127 The
unreasonableness of the burden imposed on the defendant in Robertson’s view is less a function
of the number of claims/extent of loss inflicted, and more a function of the steps required for
the defendant to fulfil his duty since Robertson focuses on liability for omissions or for acts of
third parties. There seems to be a difference between what Robertson would think of as a
burden and what Stapleton would think of as a burden which probably arises from their different
views of what the duty concept is for – since Stapleton sees it as a policy-based doctrine she
treats it as a control device, whereas Robertson treats it as an actual duty.
The community welfare concerns that Robertson then raises, which have a secondary
importance, are justiciability, community welfare, and other ‘non-justice’ factors. Justiciability
concerns judicial competence, judicial deference to legislature and executive, and cases where the
court would not be able to set a standard of care e.g. a standard owed by a pregnant woman to
her unborn child. Robertson suggests that where an issue is not justiciable this will always trump
considerations of justice between the parties. Community welfare factors include any conflict
between the duty to be owed to the claimant and duties the defendant already owes to others,
inconsistency within the legal system, availability of more effective/efficient remedies elsewhere,
effect on the community e.g. uninsurability, strain on the public purse, flood of unmeritorious
claims, inefficient allocation of risk, over-deterrence/defensive practice. Finally the ‘other nonjustice factors’ to which he refers include deontological considerations such as refusing a duty
owed by participants to one another in a joint criminal enterprise because recognition of a duty
Jane Stapleton, ‘Duty of Care Factors: a Selection from the Judicial Menus’ in Peter Cane and Jane Stapleton
(eds.) The Law of Obligations: Essays in Celebration of John Fleming (Clarendon Press 1998) 66.
127
46
would condone breaches of criminal law. Community welfare considerations alone cannot justify
liability which must be based on the demands of justice between the parties, but if there are
convincing community welfare arguments against liability then pro-duty community welfare
factors should also be weighed up.
The problem is that the introduction of these community welfare arguments introduces an
element of indeterminacy because, as we have seen, community welfare arguments introduce an
element of pluralism where no single factor has priority over the others.128 As argued earlier,
however, coherence should not be pursued to the detriment of morality, so this small element of
indeterminacy is a necessary complication. The difficulty then is in drawing a line and ensuring
that community welfare factors are only exceptionally allowed to impact on liability. Beever has
objected to policy arguments, arguing that ‘[o]ver the years, [policy] has grown from a small
suburb outlying the town of legal principles into a metropolis that now dwarfs and encroaches
upon the town.129
The ubiquity of policy, or community welfare considerations, is problematic because of the
indeterminacy inherent in it. Beever seeks to resolve this indeterminacy by adhering to a strict
corrective justice view of negligence: ‘viewing the law of negligence in terms of the principled
approach is designed to eliminate both the need for appeal to policy considerations and the
permissibility of that appeal’.130
The approach advocated in this thesis allows for consideration of community welfare concerns
but only exceptionally whilst prioritising matters of justice between the parties. To adopt
Beever’s analogy, it seeks to restore corrective justice to a metropolis and confine distributive
justice to a small suburb.
The final part of this chapter will outline the relationship between causation and the remaining
negligence doctrines. There is a danger that where causation is a problem it is thought of as
See for example Robin Oppenheim, ‘The “mosaic” of tort law: the duty of care question’ [2003] J P I Law 151,
165.
129 Beever, Rediscovering (n10) 3.
130 ibid 29.
128
47
determinative of liability. As well as considering the duty issues in the case courts must
remember that the other doctrines play their part in defining the contours of liability. It is
therefore important to know what these doctrines consist of and how they are related to
causation.
2.2
Actionable damage/the gist of the negligence action
A claimant must have suffered an actionable form of damage in order to bring a successful
negligence action; damage is said to form the gist of the action. But the question of what
constitutes actionable damage is often overlooked for the simple reason that in most cases a
claimant will have suffered clearly observable physical damage to his person or property. The
definition of damage is inextricably linked to the causation inquiry as Stapleton insists:
It cannot be over-emphasised that the formulation of the ‘damage’ forming the gist of
the action defines the causation question. Logically one can only deal with causation after
one knows what the damage forming the gist of the action is.131
Both damage and causation are then related to the later question of quantification, or valuation,
of the claimant’s loss. One theme that will recur in this thesis is that the three concepts of
damage, causation, and quantification though related must remain analytically distinct. This is
especially important where claimants have attempted to overcome difficulties of proof of
causation by reconceptualising the damage as the loss of a chance of a better outcome, or an
increase in the risk of harm.132 This section prepares the groundwork by briefly explaining the
concept of damage before outlining its relationship to causation and quantification so that the
basic issues are understood in advance of the more detailed analysis in later chapters.133
Jane Stapleton, ‘The Gist of Negligence: Part 2 The Relationship Between ‘Damage’ and Causation’ (1988) 104
LQR 389, 393.
132 E.g. Hotson v East Berkshire Health Authority [1987] AC 750, 3 WLR 232 (HL); Gregg v Scott [2005] UKHL 2, 2 AC
176; Barker v Corus [2006] UKHL 20, 2 WLR 1027. These attempts to reformulate the damage form the focus of
Chapters Three and Four.
133 Chapter Three addresses loss of a chance and Chapter Four, which addresses the evidentiary gap, considers a
range of responses including reformulating actionable damage as the risk of harm rather than the physical outcome.
131
48
2.2.1
Damage
As explained, corrective justice concerns wrongful loss in interactions. The question of whether
a loss is wrongful depends partly on whether it was caused by wrongdoing, and also on whether
it is legally recognised as constituting actionable damage. Property damage, some personal injury,
and financial loss are recognised forms of damage (although with regard to purely financial loss,
the circumstances in which one is owed a duty of care are much more limited). Conditions such
as anxiety and grief that fall short of medically recognised psychiatric illness are not actionable. 134
Similarly, not all physical illness will constitute actionable damage: it was held in Rothwell that
although asymptomatic pleural plaques are a medically recognised condition they do not
constitute actionable damage.135 Lord Hoffmann explained that damage ‘is an abstract concept of
being worse off, physically or economically, so that compensation is an appropriate remedy. It
does not mean simply a change in physical condition’.136
It is important to identify the actionable damage clearly at the outset of any claim so that the
causation inquiry can be directed to this damage. One issue that will be addressed in greater
detail in Chapter Two is the distinction between ‘divisible’ and ‘indivisible’ damage. These terms
are not used consistently amongst academics. The appropriate definition will be discussed in
detail in that chapter, but broadly speaking damage is said to be divisible if it is dose-related
whereas indivisible damage is ‘all or nothing’.
2.2.2
Quantification
At the stage of quantification of the claimant’s loss, where the task is to return him to his pretort position, causal concepts are invoked but the question is different. At the factual causation
stage of the negligence inquiry, the question is whether the defendant’s negligence was a cause of
Hinz v Berry [1970] 2 QB 40 (CA); Rothwell v Chemical and Insulating Co Ltd [2007] UKHL 39, [2008] 1 AC 281.
Rothwell (n134).
136 ibid [7] (Lord Hoffmann).
134
135
49
the damage suffered by the claimant. At the quantification stage, the court must ask whether the
claimant is any worse off because of this physical damage.
In Performance Cars v Abraham,137 the claimant’s car had been damaged in a collision due to the
defendant’s negligence. Among the damage caused to the car was damage to the front wing
which, due to the nature of the paint, necessitated the re-spraying of the whole lower part of the
car. However, in a previous collision the rear wing had been damaged which also necessitated the
re-spraying of the whole lower part of the vehicle. Although the claimant had obtained judgment
against the party responsible for the first accident, the judgment had not been satisfied and he
sought to recover the cost from the defendant responsible for the second collision. The Court of
Appeal held that although the defendant had caused physical damage to the front wing of the
car, the cost of re-spraying the lower part could not be recovered from the defendant because
the pre-tort state of the car was that the lower part already needed re-spraying. In other words,
the defendant was a cause of physical damage to the car but at the valuation stage it was found
that the physical damage had not caused the claimant to be any worse off. This is not a question
of causation – it is clear that the defendant’s negligence was a factual cause of the physical
damage – but with respect to this particular portion of the loss associated with the physical
damage the value is zero because the car already needed re-spraying before the physical damage
occurred.
At the quantification stage courts must also determine what course the claimant’s life would
otherwise have taken and whether other factors would have led to him incurring similar loss in
the future anyway. This can be illustrated through Smith v Leech Brain.138 In this case the
defendant’s negligence caused the claimant to suffer a burn on his lip which triggered a precancerous condition and led to his eventual death. As a matter of factual causation there was no
doubt that the defendant’s negligence had caused the claimant to suffer the harm at that
particular time. But in quantifying the value of the loss caused by the defendant, the court made
137
138
Performance Cars v Abraham [1962] 1 QB 33 (CA).
Smith v Leech Brain & Co [1962] 2 QB 405 (QBD).
50
a reduction in damages to account for the fact that the claimant had a pre-cancerous condition
so was likely to have died prematurely anyway. Under the ‘vicissitudes of life’ principle, courts
ordinarily must reduce the value of damages to take account of the likelihood that the claimant
would have suffered similar loss at some stage in his life, and where there is information specific
to the claimant allowing them to personalise this calculation then they will do so. This ‘reduction’
in damages does not reflect any doubt that the defendant’s negligence was a cause of the
claimant’s loss, but reflects the attempt to value that loss as accurately as possible by calculating
how likely it was that the claimant would have suffered similar misfortune at some stage in his
life because of independent events.
The question of what course the claimant’s life would have taken, and how far the defendant’s
responsibility extends, is informed by factual causation but it is also a qualitative inquiry that
involves evaluating the extent of the loss that is ‘wrongful’. This is apparent in the contrast
between the solutions in Jobling139 and Baker.140 In terms of the causal processes, Jobling and the
American case of Dillon,141 seem closer to Smith v Leech Brain than Baker does. In Jobling the
defendant caused the claimant a back injury which reduced his earning capacity but before
reaching the court the claimant suffered a naturally occurring back illness which would have
prevented him from working anyway – although the negligence did not trigger this underlying
condition as had been the case in Smith the court was still able to accurately calculate the
vicissitudes of life for this claimant by the time of trial. At the time of the negligence there was
already some idiosyncrasy of the claimant that meant that he was destined to suffer this loss.
Likewise in Dillon, as the claimant was falling to his death he was electrocuted due to the
defendant’s negligence – at the time of the negligence he already had a (significantly) reduced life
expectancy. In contrast, in Baker the claimant’s leg was damaged by the defendant’s negligence
but at that time there was no train of events already in motion that meant that he was destined to
Jobling v Associated Dairies Ltd [1982] AC 794 (HL).
Baker v Willoughby [1970] AC 467 (HL).
141 Dillon v Twin State Gas & Electric Co, 85 NH 449, 169 A 111 (1932)
139
140
51
suffer leg injury in the future. Yet by the date of trial he had been shot in that leg by robbers.
This is different from Jobling and Dillon in the sense that there was no risk of being shot particular
to the claimant at the time of the defendant’s negligence, so restoring him to his pre-tort position
is to return him to a position where he should have continued to have full use of his leg (in
Jobling/Dillon at the time of the negligence there was already a reason why the claimant would not
continue to have full use of the relevant body part in future). But this issue of quantification also
reflects the court’s qualitative assessment of the extent to which the claimant’s loss was
wrongful, with the concern in Baker that the claimant had been exposed to two careless acts and
his claim should not be allowed to ‘fall between two stools’.
2.3
Breach of duty
Just as the definition of damage is inextricably linked to the causation question, so is the framing
of the breach of duty since it must be established that the defendant’s negligence in particular,
rather than the defendant’s conduct in general, was a cause of the damage suffered. Stapleton has
illustrated this through her analysis of the decision in McWilliams v Arrol.142 The claimant fell to
his death from scaffolding and the defendant employer was negligent in failing to provide a
safety harness. To prove that failure to provide a harness was a cause of the death, it is necessary
to prove that the victim would have worn a harness if it had been provided, and the claimant was
unable to prove this. If, however, reasonable care required not only the provision of safety
harnesses but also instating a surveillance system to ensure that safety equipment was used, then
this breach of duty would be a cause of his death.143
The breach and causation inquiries are further interlinked in cases such as Bolitho where the
answer to the question of whether the damage would have occurred if the defendant had not
breached his duty depends on an assessment of the hypothetical course of action that the
142
143
McWilliams v Sir William Arrol Co Ltd [1962] 1 WLR 295 (HL).
Jane Stapleton, ‘Cause in Fact and the Scope of Liability for Consequences’ (2003) 119 LQR 388, 391-2.
52
defendant would then have followed.144 The victim in Bolitho was a child who suffered brain
damage and later died as a result of respiratory failure. He experienced two episodes of
respiratory difficulties and the doctor was negligent in failing to attend the patient when called.
The defendant argued, however, that respiratory failure would only have been avoided by
intubating the patient and if she had attended she would not have intubated, so her failure to
attend did not cause the injury. The House of Lords confirmed that courts should go on to ask
whether this hypothetical conduct would also have been negligent; a defendant cannot escape
liability by proving that that his negligence made no difference to the outcome because he would
still have caused the harm by a later act of negligence.145 This reflects the notion that the
negligence doctrines ‘articulate a single normative sequence’.146
This chapter has shown that it is important not only for the doctrine of causation to be internally
intelligible but also for it to form part of an overarching approach to the tort of negligence that is
also coherent. Aristotelian corrective justice and Kantian Right, it was argued, provide a coherent
theoretical foundation that is not only formal but also implies a moral account of wrongdoing
that expresses interpersonal responsibility. The various negligence doctrines should come
together to give effect to this corrective justice-based interpersonal responsibility. The duty of
care is central to the negligence inquiry; it is at the duty stage that the courts will identify the
issues of interpersonal responsibility that should shape application of the remaining doctrines so
that they ‘articulate a single normative sequence’. Causation is an essential element of
interpersonal responsibility because it is the causal relationship between the wrongdoing and the
loss that joins the defendant and claimant in an interaction that has resulted in a ‘wrongful loss’.
Bolitho v City of Hackney Health Authority [1998] AC 232 (HL).
The Court of Appeal later clarified in Gouldsmith v Mid-Staffordshire General Hospitals Trust [2007] EWCA Civ 397,
[2007] LS Law Medical 363, that there are two stages to this inquiry: first, what hypothetical course of action would
the defendant/third party probably have taken?; second, if they would not have given harm-preventing treatment
would this have been negligent? If the answer to the first question is that the defendant/third party probably would
have undertaken treatment to prevent the harmful outcome then it is clear that their initial negligence did cause the
harm and it is not necessary to prove that they would have been bound, in the Bolam/Bolitho sense, to follow this
hypothetical course of action.
146 Weinrib, ‘Legal Formalism’ (n55) 593.
144
145
53
However, causation also has a limited role and the evaluative elements of responsibility must be
addressed through the remaining negligence doctrines. These doctrines also help frame the
causation question because they identify the loss and the wrongdoing between which the causal
relationship is sought.
This foundation means that the analysis of causation that will be developed in the remaining
chapters is based on a clear understanding of the underlying theory of negligence law and of the
demands that negligence makes of the doctrine of causation.
54
Chapter Two: Identifying the Proper Function of Causation
Introduction
As established in chapter one, causation is central to interpersonal responsibility, but the law
relating to causation is currently in a state of confusion. This confusion extends not only to the
exceptional approaches to causation, but characterises the standard approaches as well. This
chapter will attempt to clear away the confusion by adopting a clear approach to understanding
the causal problems as well as their solutions. The starting point to developing any test for
causation must be an accurate understanding of what purpose the doctrine of causation has in
negligence. It will be argued here that the concept of causation in law is narrower and more
precise than the use made of the term ‘cause’ in everyday language because, as Stapleton has
explained, causation can be understood in a wide range of senses, from involvement, to
explanation, to blame.1 Moreover, as Wright, building on the work of Hart and Honoré, argues,
philosophy provides the law with a robust account of causation that ought to form the basis of
any legal test.2 By understanding the philosophical account of causation developed by Hume and
Mill, we can visualise and better understand the causal problems that come before the courts.
This will enable the law to distinguish between those cases that are straightforward, and those
that are actually problematic. Negligence law does, however, put its understanding of causation
to a specific purpose, namely establishing a causal link between a particular instance of
negligence and a particular harmful outcome.3 A certain level of pragmatism must therefore be
allowed in order to avoid becoming distracted by details that affect philosophers and scientists.
Effectively, we all use a shared concept of causation, derived from philosophy, but we must
retain a clear vision of what demands the law places on a test of causation.
Jane Stapleton, ‘Choosing what we mean by “Causation” in the Law’ (2008) 73 Mo L Rev 433, 438.
Richard Wright, ‘Causation in Tort Law’ (1985) 73 Cal L Rev 1735 (hereafter ‘Causation’), 1774-1812.
3 Richard Wright, ‘The NESS Account of Natural Causation: A Response to Criticisms’ in Richard Goldberg (ed.)
Perspectives on Causation (Hart Publishing 2011) 290-1.
1
2
55
The philosophical account of causation is best reflected by Richard Wright’s NESS test rather
than the but-for test that currently dominates.4 The but-for test is mismatched with the
philosophical reality of what causation is and this means that the use of the but-for test has
created apparent problems that do not, in reality, exist. The legal solutions to these problems,
particularly the ‘material contribution to harm’ test, have created more difficulties than they have
solved. The reasons for this are twofold. First, because causation is approached from the
paradigmatic mind-set of the but-for test which is not an accurate reflection of the meaning of
causation, we are unable to conceptualise the problems that exist. This has contributed to the
second difficulty regarding the emergence of a vocabulary of causation-specific terminology:
divisibility of disease, cumulative causes, alternative causes. These terms are technical so they
ought to have a technical meaning but the difficulties of conceptualising the causal problems to
which they relate means that they do not have any agreed meaning. They are used to mean
different things by different people, thus exacerbating the confusion. By beginning again from
the most basic causal problems and explaining them through the conceptual framework of the
NESS test, many of these problems can be avoided.
This chapter will therefore begin by asking how the legal concept of causation relates to
causation in the fields of everyday language, and philosophy and science. After explaining the
philosophical account of causation, it addresses the question of how best to translate this into a
legal test. Both the but-for test and the NESS test will be applied to a range of theoretical
problem scenarios to illustrate how the NESS test is able to solve problems that the but-for test
cannot. Finally, the NESS test will be applied to existing case law to show how it allows us to
reconceptualise the problems and to understand the specific terminology.
Wright, ‘Causation’ (n2); Richard Wright, ‘Causation, Responsibility, Risk, Probability, Naked Statistics, and Proof:
Pruning the Bramble Bush by Clarifying the Concepts’ (1987) 73 Iowa L Rev 1001 (hereafter ‘Pruning’); Wright,
‘The NESS Account’ (n3). Wright builds on the idea of a ‘causally relevant factor’ first developed by Hart and
Honoré, Causation in the Law (Clarendon Press 1959).
4
56
1. The demands on causation
Before it is possible to say what qualities are required from a test for causation, we must
understand what is actually meant by ‘causation’. More importantly, we need to know what we,
as lawyers, mean by causation. What role does it fulfil in the negligence inquiry, and does it have
a special ‘legal’ definition? In the previous chapter it was shown that causation has a pivotal yet
limited place in corrective justice, and the two strands of the current section will unpack what
this implies for the doctrine of causation within the negligence inquiry. First, causation has a
limited role in attributions of responsibility so as Wright argues, in comparison with the use of
the word ‘causation’ in everyday language, there is a technical legal meaning.5 ‘Causation’ in
everyday language is used flexibly to mean different things in different circumstances. In
contrast, the law requires greater consistency and precision in the definition of causation. The
factual causation element of negligence is concerned with causation in a purely factual sense,
what Hart and Honoré call being a ‘causally relevant condition’, or Stapleton calls ‘involvement’.
The second part of this section addresses the point that causation is essential to interpersonal
responsibility meaning that the doctrine of causation must be conceptually robust. As argued
first by Hart and Honoré, and subsequently by Wright, the most complete account of factual
causation can be found in the philosophy of Hume and Mill, and this should inform the legal
doctrine of factual causation.
1.1
The limited role of causation in negligence
It has been suggested that the doctrine of causation should reflect the ordinary man’s use of the
word, for example in McGhee v National Coal Board, Lord Reid said ‘it has often been said that the
legal concept of causation is not based on logic or philosophy. It is based in the practical way in
Indeed Wright suggests that although Hart and Honoré’s philosophical account of causation was ‘a major advance
in the analysis of causation in both law and philosophy’ it was ‘overshadowed and distorted by their primary
emphasis on elaborating supposedly factual ‘common sense’ principles for treating only some causally relevant
factors as causes, so that, initially, it received minimal attention in the legal literature’. See Wright ‘The NESS
Account’ (n3) 288.
5
57
which the ordinary man’s mind works in the everyday affairs of life’.6 Indeed the seminal work of
Hart and Honoré on causation sought to explain causation by reference to the way it is used in
ordinary language.7 Wright has shown, however, that in everyday language ‘cause’ is often
shorthand for ‘the cause’ rather than ‘a cause’:
The phrase “a cause” usually refers to causation per se – the fact of being one of many
contributing conditions. The phrase “the cause” generally is used to denote which of the
many contributing conditions is legally or morally responsible…“The cause” is merely an
elliptical way of saying “the (most significant for our purposes) cause”.8
In other words the ordinary man uses the word ‘cause’ to express a conclusion about
responsibility or blame rather than limiting himself to observing a purely factual relationship of
cause and effect. In comparison, the legal notion of factual causation is a technical concept since
it refers to ‘a cause’ so care must be taken to separate the fact of causation from any attribution
of responsibility that may be considered to flow from it.
A relatively straightforward example illustrates the distinction Wright draws.9 When a lit match is
dropped and a fire results, the ordinary man might identify the dropping of the match as the
cause. It is, however, only ‘a cause’, along with the presence of oxygen and flammable material. If
a fire occurred during a particular manufacturing process that required the absence of oxygen,
then common sense is to call the introduction of oxygen the cause of the fire. If a sofa does not
meet safety standards concerning flammability then we might say that this, not the dropping of a
match onto it, is the cause of the fire. In each of these scenarios ‘the cause’ is selected because it
is the factor that is abnormal in the circumstances, and as Mackie has argued, ‘[w]hat is normal
may depend upon man-made norms’.10 The identification of ‘the cause’ depends upon the
purpose of the inquiry and the circumstances of the event. When the ordinary man determines
McGhee v National Coal Board [1972] 3 All E.R. 1008 (HL), 1011.
Hart and Honoré, Causation in the Law (2nd ed., Clarendon Press 1985).
8 Wright, ‘Pruning’ (n4) 1012.
9 This example is used by Hart and Honoré (n7) 11.
10 J. L. Mackie, The Cement of the Universe: A Study of Causation (Clarendon Press 1974) 119.
6
7
58
the cause of an event, such as a fire, he goes through two stages, the first factual and the second
evaluative. The second of these stages is to pick out the particular factor that was unusual or
blameworthy in the circumstances, which in the case of the fire above could be the match or the
presence of oxygen or the flammability of the material depending on the circumstances. But first,
although he did not necessarily consciously articulate it, he was aware that the conditions that
must come together for a fire to start are oxygen, flammable material and a source of ignition.
Each of these conditions is ‘a cause’ of a fire because they must each be present, but the ordinary
man tends only to identify one as ‘the cause’, and which one he chooses will vary depending on
the particular circumstances. As a test of factual causation it is therefore inappropriate for the
law to rely on the ordinary man’s use of the word causation because the ordinary man makes an
evaluative judgment without first articulating the factual judgment on which it is based.
The distinction between ‘a cause’ and ‘the (responsible) cause’ broadly corresponds to the
division between factual and legal causation in negligence. Factual causation is truly causal, legal
causation is an evaluative question of the appropriate extent of liability. Hart and Honoré,
however, identified three issues and it is important to understand which of these are causal and
which are evaluative. These three issues are: causally relevant condition, causal connection, and
remoteness of damage.11 The identification of ‘causally relevant conditions’ is the factual
question of the historical involvement of the negligence in the occurrence of the harm.12 This is
what Wright referred to above as being ‘a cause’. The second question of ‘causal connection’
identifies which of the causally relevant conditions actually are, in Hart and Honoré’s
terminology, causes and which are just conditions.13 It was in making the distinction between
causes and conditions that Hart and Honoré turned to the use of the word ‘cause’ in everyday
language. They argued that the ordinary man identifies something as a cause rather than a
condition if it is non-coincidental, is free and deliberate human action or if it is something
Hart and Honoré (n7) xlvii.
ibid lvi, lxiv.
13 ibid 33.
11
12
59
abnormal. Finally the third issue considers whether the harm was too remote a consequence of
the defendant’s conduct. This, Hart and Honoré conceded, is a policy-based decision as to the
appropriate scope of liability.14 In essence they identified the first stage as factual, the second as
being based on principle and the third as based on policy.
For them, the first two issues are truly causal and the final issue is non-causal. This conclusion
was probably driven by Hart and Honoré’s aim of countering the Legal Realists’ criticisms that
causation is not neutral but is policy-driven.15 Hart and Honoré’s work is important, among
other reasons, for having shown that the task of identifying ‘the cause’ from among everything
that could be called ‘a cause’ is not driven purely by policy but also by principle. However
although the identification of causal connection is based on principle it is still an evaluative
process and as such it ought not be considered part of factual causation.
As Stapleton has highlighted, when Hart and Honoré address ‘causally relevant condition’ they
say that ‘a cause is basically like an element in a recipe’ but when they discuss ‘causal connection’
they say that ‘a cause is…an intervention in the existing or expected course of events’. 16 To say
that both of these inquiries are causal would create confusion whenever the word ‘cause’ is used
as to which of these senses is intended. As Stapleton argues, ‘[n]ot only are there two distinct
underlying enquiries, one historical and one purposive, but the disputes to which they give rise
are of a fundamentally different nature’.17 Glanville Williams explained that Hart and Honoré’s
contention that ‘ “it is the plain man’s notions of causation (and not the philosopher’s or the
scientist’s) with which the law is concerned”…may be partly true of the legal notion of
proximity, but it is not true of the lawyer’s use of the notion of but-for causation’.18The first
stage of identifying ‘causally relevant conditions’ is the factual aspect of causation which
identifies the defendant’s negligence as a cause and it is from this that Wright later developed the
ibid 305-307.
Wright, ‘Causation’ (n2) 1739.
16 Hart and Honoré (n7) 31 and 29. See Jane Stapleton, ‘Unpacking Causation’ in Peter Cane and John Gardner
(eds) Relating to Responsibility: Essays in Honour of Tony Honoré on his 80th birthday (Hart Publishing 2001) 159.
17 Stapleton (n16) 159.
18 Glanville Williams, ‘Causation in the Law’ [1961] CLJ 62, 65.
14
15
60
NESS account of causation. It is for this stage that the label of ‘factual causation’ ought to be
reserved.19 The later stages of causal connection and remoteness are both evaluative and address
the question of whether the defendant can be considered the cause and should be grouped
together under the rubric of ‘legal causation’.
Stapleton has highlighted an advantage of limiting the doctrine of causation to the purely factual
question of involvement:
This would then require us to locate the normative controversies about their different
degrees of responsibility for the death under analytical labels such as “duty,”
“breach,”…and so on. The great attraction of this approach is that under these analytical
labels, unlike the label of “causation,” it has traditionally been unacceptable merely to
assert a conclusion on the basis of “intuition” or “common sense.”20
So although the notion of a causally relevant condition is very inclusive when viewed in isolation,
causation alone is not determinative of liability and the other doctrines enable the law to narrow
liability from causal responsibility to interpersonal responsibility.
1.2
The robustness of the philosophical account
The NESS test for factual causation that was developed by Wright, building on the work of Hart
and Honoré on causally relevant conditions, is based on the dominant philosophical account of
causation. This is David Hume’s regularity account of causation and counterfactual analysis, later
modified by John Stuart Mill.
Lord Hoffmann has questioned why academics insist on strict factual causation and say that if a
court holds something that ‘did not qualify as ‘cause in fact’ as nevertheless satisfying its causal
Stapleton has argued that since ‘legal causation’ is evaluative and not truly a question of causation it should be
relabelled ‘scope of liability for consequences’ to avoid confusion (‘Cause-in-Fact and the Scope of Liability for
Consequences’ (2003) 119 LQR 388). Analysis of this proposal is outside the scope of this thesis, what matters is
that the doctrine of ‘factual causation’ has been defined.
20 Stapleton, ‘Choosing’ (n1) 446.
19
61
requirements, then it should be regarded as deeming something to be a cause when it was not
really a cause’.21 He argues:
It is this concept of something having to be really a cause according to criteria lying
outside the law which puzzles lawyers. On what basis are academic writers entitled to say
that judges should take into account a philosophically privileged form of causation which
satisfies criteria not required by the law?22
However, given that causation is essential to corrective justice it is important that the concept of
causation adopted in law reflects the natural relation. It is the task of philosophers of causation
to understand and explain this, rather than it being a legal creation or fiction. As Moore explains:
[C]orrective-justice…demands a robustly metaphysical interpretation of cause. For legal
liability tracks moral responsibility on this view, and moral responsibility is for those
harms we cause. ‘Cause’ has to mean what we mean when we assign moral responsibility
for some harm, and what we mean in morality is to name a causal relation that is natural
and not of the law’s creation.23
Seeking an explanation from philosophy has been called ‘philosophically naïve’ by Broadbent
who accuses Richard Wright of ‘underestimat[ing] how difficult it is to provide a satisfactory
account of causation itself’.24 As Fumerton and Kress have also observed, ‘[p]hilosophers have
labored long and hard on the question of how to analyze causation, with a striking lack of
success’.25 Furthermore, in the absence of a ‘fully adequate universal test for factual causation’,
Broadbent asserts that there can be no such thing as ‘factual causation’ and that it will always be
determined by the law.26 However while the philosophical account may remain incomplete in
places, it is the most complete account of causation available so it provides the law with the most
Lord Hoffmann, ‘Causation’ in Richard Goldberg (ed.) Perspectives on Causation (Hart Publishing 2011) 5.
ibid.
23 Michael S. Moore, Causation and Responsibility: An Essay in Law, Morals, and Metaphysics (Oxford University Press
2009) 95.
24 Alex Broadbent, ‘Fact and Law in the Causal Inquiry’ (2009) 15 Legal Theory 173, 176.
25 Richard Fumerton and Ken Kress, ‘Causation and the Law: Preemption, Lawful Sufficiency, and Causal
Sufficiency’ (2001) 64 Law and Contemp Prob 83, 102.
26 Broadbent (n24) 178.
21
22
62
robust account of causation that is currently possible. Pragmatism is needed in law because the
need to reach concrete decisions precludes waiting for philosophical certainty. Miller explains,
‘[p]oliticians are not philosophers; they cannot wait for certainty before, for example, banning a
drug which appears to have a harmful side-effect. Certainty will be similarly absent in evidence
presented in any civil action taken by those so harmed’.27 While there do remain some points of
philosophical debate about the meaning of causation, there is a dominant account and it is
preferable for the law to adopt this account rather than seeking a distinct definition.
Having separated issues of responsibility, principled and policy-based, from issues of causation,
it is now necessary to consider exactly what causation is. What does it mean to be a factual cause,
or in Hart and Honoré’s classification a ‘causally relevant condition’?
2. The philosophical account of causation
To understand what it means to be a cause Hart and Honoré, and subsequently Wright, turned
to the dominant philosophical account of causation provided by Hume and Mill.
2.1
Hume: contiguity, priority, constant conjunction and necessity
According to Hume’s account of causation the qualities that can be observed in the relationship
between causes and effects are contiguity, succession and constant conjunction. These
observations form the regularity account of causation. They can be illustrated with an example
provided by Hume:
Here is a billiard-ball lying on the table, and another ball moving towards it with rapidity.
They strike; and the ball, which was formerly at rest, now acquires a motion…There was
no interval betwixt the shock and the motion. Contiguity in time and place is therefore a
requisite circumstance to the operation of all causes. ‘Tis evident likewise, that the
motion, which was the cause, is prior to the motion, which was the effect. Priority in time,
27
Chris Miller, ‘Causation in personal injury: legal or epidemiological common sense?’ (2006) 26 LS 544, 547.
63
is therefore another requisite circumstance in every cause. But this is not all. Let us try
any other balls of the same kind in a like situation, and we shall always find, that the
impulse of the one produces motion in the other. Here, therefore, is a third circumstance,
viz. that of a constant conjunction betwixt the cause and effect. Every object like the cause,
produces always some object like the effect.28
Hume then went on to consider the question of whether there is a relationship of necessity
between cause and effect, and this is where the novelty of his account lay as Wright explains:
Hume revolutionized philosophic thinking on causation when he insisted that, contrary
to the then-popular belief, singular causal judgments are not based on direct perception
of causal qualities or forces inherent in objects or events: no such quality or force has
ever been identified. Instead, causal judgments are based on the belief that a certain
succession of events instantiates one or more causal laws.29
In other words, we observe that necessity is what distinguishes a causal relationship but it does
not exist as a quality internal to objects. So for any relationship of objects displaying contiguity,
succession and constant conjunction, the element that distinguishes the causal from the noncausal is necessity. This introduces the element of counterfactual analysis in causation that allows
us to test whether a factor that satisfies the regularity account is actually a cause. For example, an
‘epiphenomenon’ displays contiguity, succession and constant conjunction but does not cause
the outcome: if c is an epiphenomenon of the causal history of e it is ‘a more or less inefficacious
effect of some genuine cause of e’.30 An illustration of this is the needle on a barometer dropping
before a storm. The drop in pressure c, first causes the barometer needle to fall effect e, and then
causes a storm effect f: ‘c causes first e and then f, but e does not cause f’.31 If the barometer is
David Hume, An Abstract of a Treatise of Human Nature, ed. JM Keynes and P Sraffa (Cambridge University Press,
Cambridge 1938) 11f.
29 Wright, ‘Pruning’ (n4) 1019.
30 David Lewis, ‘Causation’ (1973) 70 Journal of Philosophy 556, 557.
31 ibid 566.
28
64
broken and the needle does not fall this does not affect the fact that the drop in the pressure has
occurred so effect f, the storm, will still occur.
The necessity requirement explains an aspect of the House of Lords’ decision in Rothwell v
Chemical and Insulating Co Ltd32 concerning claims for pleural plaques. Among other heads of
damages, the claimants were unable to recover for the increased risk of contracting asbestosrelated illness because pleural plaques do not increase this risk. Pleural plaques will often be
observed prior to the development of other asbestos-related illness, however they do not
increase the risk of such illness but merely signal that the patient is at an increased risk of
contracting an illness because he has inhaled asbestos dust. It is the quality of necessity that
differentiates asbestos dust as causative of certain cancers and pleural plaques as non-causative
but merely indicative of the risk. Pleural plaques are, in effect, a human equivalent of the needle
on a barometer dropping before a storm.
In summary, causation thus consists of three empirical elements: contiguity, succession and
constant conjunction; and a non-empirical element: belief in necessary connection.33 Despite its
non-empirical character, Hart and Honoré have observed that necessity has entered modern
thinking in an altered form having ‘changed its psychological form for a logical one’.34
2.2
Mill: the addition of ‘plurality’ and ‘complexity’
Hume, however, still largely wrote about ‘an object’ or ‘an event’ as cause and effect, so of a
cause having a single effect, and an effect having a single cause. Mill subsequently made two
crucial modifications to Hume’s account of causation, recognising the ‘plurality of causes’ and
the ‘complexity of causes’.35
Rothwell v Chemical and Insulating Co Ltd and another, Topping v Benchtown Ltd (formerly Jones Bros Preston Ltd), Johnston v
NEI International Combustion Ltd, Grieves v F T Everard & sons Ltd and another [2007] UKHL 39, [2008] 1 AC 281.
33 Tom L Beauchamp & Alexander Rosenberg, Hume and the Problem of Causation (Oxford University Press 1981) 4.
34 Hart and Honoré (n7) 14.
35 Hart and Honoré (n7) 19. Writing about Mill, System of Logic.
32
65
2.2.1
Complexity of causes
The complexity of causes means that an occurrence does not have a single object or event as its
cause, but the cause is the sum total of a set of positive and negative conditions. Drawing on the
scenario used by Hume above, the causal law concerning the movement of the billiard ball did
not just contain the event of being struck by the other ball, but of being struck by a ball ‘of the
same kind in a like situation’. Being struck by a smaller ball, at a different speed, or on a different
surface, may not lead to the same result. So an event is caused by the coming together of a set of
conditions. Wright explains:
A fully specified causal law or generalization would state an invariable connection
between the cause and the consequence: given the actual existence of the fully specified
set of antecedent conditions, the consequence must follow.36
Each of the conditions that makes up the complete set has an equal status. Each is equally
deserving of being called a cause, because in the absence of any of them the outcome would not
have occurred: ‘it is in fact related to the effect in precisely the same way as the other
constituents of the set’.37 The idea that every factor that makes up the set is equally deserving of
the label ‘cause’ is a situation Mackie describes as being ‘symmetrical with respect to all the
factors’.38 Causation, in Honoré’s words, provides a ‘recipe’ for the set of conditions that are
together sufficient to produce a given result.39
As noted in section 1.1, the causal inquiry in negligence is focused on the question of whether
the negligence was a cause of the damage, so it is unnecessary to provide the whole ‘recipe’. This
does not prevent the legal concept of causation from being based on the philosophical account
since it is possible to remain faithful to Mill’s explanation of causation whilst on a practical level
Wright, ‘Causation’ (n2) 1789.
Hart and Honoré (n7) 21.
38 Mackie (n10) 34.
39 Tony Honoré, ‘Necessary and Sufficient Conditions in Tort Law’ in David G Owen (ed.) Philosophical Foundations of
Tort Law (Oxford University Press 1995) 375.
36
37
66
being content to only partially populate the sufficient set that instantiates the causal law in
question as Wright explains:
In the typical singular causal statement, the causal assertion includes, explicitly or
implicitly, only a few of the antecedent conditions but nevertheless asserts that they were
only part of an incompletely specified (and incompletely understood or known) set of
actual conditions that was sufficient for the occurrence of the consequence.40
Beever criticises this position, arguing that it means that ‘the law’s understanding of causation is
looser than science’s’ and that this is inconsistent with the requirement that ‘the lawyer’s
conception of causation must be philosophical and scientific for it to be a conception of
causation’.41 Beever provides an example of a barn that is burnt down because the defendant
placed his rick of hay against it. He says that science would require not just the presence of the
rick of hay, but all the other conditions making up the sufficiency of the set for the causal law,
such as the dryness of the hay, sufficient heat, the barn being made of flammable material, the
presence of oxygen etc. In contrast, Wright would say that these conditions are not relevant in
law, because the law is satisfied to say that the rick of hay was one antecedent condition which
was part of an incompletely specified set. This leads Beever to believe that Wright is saying that
the legal concept of causation is different to the scientific concept of causation. To support this
argument he says that the other constituents of the set are not irrelevant because the court would
not be likely to conclude that the rick of hay caused the destruction of the barn if the rick was
soaking wet, or the hay did not ignite, or the barn was made of concrete. Whilst Beever is right
to say that the court would reach a different conclusion about causation if it knew that any of
these other conditions were present, this does not mean that it adopts a different, unscientific
and unphilosophical concept of causation if it does not list all of the conditions. The underlying
concept of causation is the same as the philosophical or scientific one, but the court’s purpose is
not to discover every element of the causal set. Instead the court’s purpose is to determine
40
41
Wright ‘Causation’ (n2) 1789.
Allan Beever ‘Cause-in-fact: Two Steps Out of the Mire’ (2001) 51 UTLJ 327, 344.
67
simply whether one specific candidate condition, the defendant’s negligence, in fact belonged to
the set. The court therefore relies on the information presented to it, so if the defendant
established that counter-conditions, such as the rick being soaking wet, had been present, then
the court will add these into the causal set. But if no such conditions are presented, then as a
practical measure the court is justified in being satisfied that the rick was a cause of the
destruction of the claimant’s barn. The concept of causation does not change, but pragmatism
leads the law to be satisfied to address only those candidate conditions that are presented to it.
2.2.2
Plurality of causes
In addition to causes being ‘complex’, they are also ‘plural’: there is not just one set of conditions
that is capable of causing the outcome, instead there are a number of different possible sets, each
of which is capable of causing the same outcome. So, for example, a billiard ball may move
because it is struck by another ball, or because the table is tilted or because air is blown at it etc.
In summary, a single outcome is caused by a set of conditions rather than having a single cause,
and various different sets of conditions can exist that are all capable of causing the same single
outcome.
In this account necessity is still essential as the quality that distinguishes causally relevant
antecedents:
[O]ne must be careful when constructing a causal law or generalization to distinguish the
causally relevant antecedent conditions from the causally irrelevant antecedent
conditions. This differentiation is necessary to insure that the set of jointly sufficient
antecedent conditions includes only those that are indeed invariably connected with the
consequence. Thus, the antecedent conditions must be restricted to those that are
necessary for the sufficiency of the set.42
42
Wright, ‘Causation’ (n2) 1790.
68
Necessity has been confined to a smaller role because for a given effect it is not necessary for it to
be preceded by a particular cause since there exist a number of possible sets capable of causing
the effect. So although each condition is necessary for the sufficiency of a particular set, each
particular set can only be described as ‘sufficient’ to bring about the effect.
Hart and Honoré noted a philosophical objection to the idea of plurality of causes in the
argument that an equal plurality of effects must exist but we fail to differentiate effects with the
same precision that we apply to causes: the ‘apparent plurality of causes is due to our failure or
inability to analyse the effects with the same particularity as the causes’.43 If we were able to
particularise any single effect sufficiently we would see that it only has one possible cause. This is
similar to saying that we must be more precise than simply saying a person died and, for
example, specify that he died from a heart attack, or from poisoning. It is simply saying that we
need to continue this process of precision in our description of the event until we have as many
effects as we have causes.
They noted that this philosophical objection is not universally accepted, but argued that whether
or not it is well-founded, ‘the lawyer, the historian, and the plain man accept the doctrine that an
event of a given kind may be produced by different causes’.44 That the doctrine of plurality of
causes is accepted can be illustrated by close consideration, for example, of the cause of a heart
attack. A heart attack occurs because part of the heart muscle dies when it is starved of oxygen,
usually because of a blockage in a coronary artery. Such blockages can occur for a multitude of
reasons with risk factors including smoking, high blood pressure, lack of exercise, and poor
diet.45 Rather than attempting to analyse the effect, the heart attack, in ever more detail, the
ordinary man accepts that the effect has a plurality of possible causal sets, each containing one of
the many possible combinations of these, and other, risk factors. Hart and Honoré’s pragmatism
is justifiable on the basis that if we did not accept plurality of causes then we would never accept
Hart and Honoré (n7) 20.
Hart and Honoré (n7) 20.
45 http://www.nhs.uk/Conditions/Heart-attack/Pages/Causes.aspx Accessed 14th January 2010.
43
44
69
any causal law as being true, and never be able to make causal judgments. Causal conclusions
cannot wait until the moment when, indeed if, science advances to a point where effects can be
adequately particularised. In other words, this is one point where pragmatism is necessary since
the law cannot wait for philosophy and science to become settled.46
3. Tests for causation
So far this chapter has identified the demands that negligence places on the doctrine of
causation, and the philosophical account of the concept of causation. It is now possible to build
on these foundations to analyse the possible tests for causation. It will be argued that the NESS
test should be adopted instead of the but-for test because it reflects the concept of causation
more accurately. While the but-for test has the attraction of apparent simplicity compared to the
NESS test, it only functions correctly in straightforward cases, and in those cases the NESS test
operates just as simply. In the more complex cases of over-determined causation, the but-for test
fails because it is not an accurate test of causation, so courts have developed exceptions whose
scope and meaning are unclear. In contrast, the NESS test is able to cope with these more
complex cases, and because it accurately reflects the concept of causation it enables us to
visualise the causal problems and solutions without resorting to exceptional tests.
3.1
Complexity and plurality of causes as necessary and sufficient conditions
The philosophical account of causation described above established the following qualities as
characteristics of causes: contiguity of cause and effect, priority in time and constant
conjunction. In addition, necessity is the quality that distinguishes causal factors from non-causal
factors. Moreover, causes are complex: an outcome is the result of a set of conditions coming
together; and plural: multiple possible sets can each bring about the same outcome. The NESS
46
See text to (n27).
70
test will be shown to be preferable to the but-for test on a theoretical level because it more
accurately reflects these qualities.
There are different degrees to which a single factor, such as a defendant’s negligence, might be
thought sufficient or necessary for the occurrence of the harm. These are explained by Wright:
In descending order of stringency, a strict-necessity test requires that Q be necessary for
the occurrence of R whenever R occurs; a less stringent, strong-necessity test requires
only that Q have been necessary for the occurrence of R on the particular occasion,
considering the circumstances that existed on the particular occasion; and the least
stringent, weak-necessity test requires only that Q have been a necessary element of some
set of actual conditions that was sufficient for the occurrence of R… A strict sufficiency
test requires that Q be sufficient by itself for the occurrence of R; a less stringent, strongsufficiency test requires only that Q be a necessary element of some set of existing
conditions that was sufficient for the occurrence of R…; and the least stringent, weaksufficiency test “requires” only that Q be a part of some set of existing conditions that
was sufficient for the occurrence of R.47
Some of these types of test can easily be discarded because they clearly do not reflect the
characteristics of causes. Strict necessity would mean that a particular effect only ever has one
possible cause, but this is precluded by the notion of plurality of causes. Similarly, strict
sufficiency, the idea that one single factor can be sufficient to bring about an effect, is precluded
since causes are ‘complex’. Weak sufficiency does not incorporate the vital characteristic of
necessity to any degree so it is unable to determine whether a factor was a cause. The remaining
possibilities are strong necessity and weak necessity (or its equivalent, strong sufficiency). The
but-for test asks whether a factor was necessary for the occurrence of harm on the particular
occasion so it is a test of strong necessity. The NESS test does not require the factor to have
been necessary for the harm on the particular occasion, but allows for the possibility that more
47
Wright ‘Pruning’ (n4) 1020.
71
than one sufficient set of factors may have existed on the particular occasion. As long as the
factor was necessary for the sufficiency of one of the sets that actually occurred on the particular
occasion then it was a cause. The NESS test therefore corresponds to weak necessity. 48 Although
the but-for test acknowledges the plurality of causes insofar as it recognises that a variety of
different sets could potentially cause a particular outcome, it does not take the notion of plurality
to its logical conclusion because it assumes that each potential set only ever occurs singularly.49
The NESS test recognises that more than one set may actually occur simultaneously. Practical
examples will show that this complete incorporation of plurality is correct.
3.2
But-for and NESS in simple cases
The but-for test is illustrated clearly by the case of Barnett v Chelsea and Kensington Hospital
Management Committee50 where the claimant’s husband became ill after drinking a cup of tea. He
went to hospital where the staff were negligent in their failure to examine and treat him, instead
sending him home with advice to see his GP the next day. The victim died from arsenic
poisoning in his tea but despite the negligence of the hospital staff his claim failed at the
causation stage because but-for their negligence he would still have died. Given the time that had
elapsed between drinking the tea and arriving at hospital, the time it would have taken to
examine and diagnose the illness and begin treatment, the treatment would have failed to save
him. Given the circumstances, including the stage of advancement of the poisoning, the doctor’s
negligence was not necessary for the outcome, it made no difference. Herein lies the appeal of
the but-for test, it answers the question ‘did the defendant’s act make a difference?’.51 Yet it is
widely recognised that the but-for test is of limited use because as well as excluding irrelevant
Wright, ‘Pruning’ (n4) 1021
Mackie’s INUS test (‘insufficient but necessary part of an unnecessary but sufficient set’ test) incorporates the
plurality of causes since it recognises that a particular sufficient set of conditions need not be necessary for the
outcome. See JL Mackie, The Cement of the Universe: A Study of Causation (Clarendon Press 1974). For any particular
instance of causation, however, it only recognises the occurrence of one sufficient set which, as Wright explains,
converts the INUS account into the sine qua non (‘but-for’) account. See Wright, ‘The NESS Account’ (n3) 288. Since
it is applied in the same way as the but-for test it will not be analysed separately.
50 [1968] 2 WLR 422 (QBD).
51 Moore (n23) 84.
48
49
72
conditions ‘it can also exclude others that are relevant’52 in cases where an outcome is ‘overdetermined’ as in the classic double-hit hunters scenario discussed in the next section.
The NESS test was developed by Richard Wright in response to the weaknesses of the but-for
test and built upon ideas initially developed by Hart and Honoré in Causation in the Law. In this
account of causation something is a cause if it was a Necessary Element of a Sufficient Set: ‘a
particular condition was a cause of (condition contributing to) a specific consequence if and only if it was a
necessary element of a set of antecedent actual conditions that was sufficient for the occurrence of the consequence.
(Note that the phrase “a set” permits a plurality of sufficient sets.)’.53 The first part of this is
uncontroversial, since causes are ‘complex’ any outcome is caused by the coming together of a
set of conditions which are together sufficient to cause it. The novelty of the NESS test is that it
fully incorporates plurality, so while in most instances one particular outcome will only have
been preceded by one of these possible sets, the NESS test recognises that multiple sets may
actually occur.
Where only one set actually occurs the NESS test would not complicate the way the law
currently works. For example in Barnett there are many ways that the claimant’s husband’s death
could have been caused but only one, poisoning with arsenic, actually occurred. The doctor’s
negligence was not necessary for the sufficiency of the set of conditions that actually occurred
because the treatment would still have been too late even if the doctor had examined the patient.
In these straightforward cases the NESS test would effectively collapse into the but-for test; the
but-for test is shorthand for the NESS test where there is only one set of conditions. This makes
it clear that widespread adoption of the NESS test would not have the effect of complicating
cases generally which should allay possible fears that adopting the NESS test would lead to
increased litigation.
52
53
Honoré (n39) 363-4.
Wright ‘Causation’ (n2) 1790.
73
Since strong necessity (but-for) is more stringent, a factor that is strongly necessary will also
satisfy a test of weak-necessity. But the reverse is not true – a factor that fails the test of strongnecessity may still be a cause because causation only requires weak-necessity. The but-for test can
tell us that something was a cause, it is ‘a test of inclusion’,54 but it cannot accurately tell us
whether something was not a cause. As Honoré has said:
No one will deny that the but-for test has in many instances a heuristic value: it often
provides a quick way of testing the existence of causal connection. However, it is another
matter whether it is part of the meaning of ‘causally relevant condition’ or ‘cause’.55
The but-for test fails to deal adequately with more complex causal problems because it does not
accurately correspond to the meaning of causation. So the fact that we are often able to rely on
the but-for test should not be allowed to mislead into the belief that it is the best test to use. As
Wright argues, ‘the substitute must give way to the more accurate and comprehensive concept
when the situation is more subtle and complex’.56
3.3
Over-determined causation
‘Over-determination’ means that there seem to be too many causes. Where this takes the form of
‘duplication’ two or more possible causes occur at the same time, and in ‘pre-emption’ the
possible causes occur one after the other.57
Wright, ‘Pruning’ (n4) 1022.
Honoré (n39) 367.
56 Wright, ‘Causation’ (n2) 1792.
57 Hart and Honoré used different labels to describe the forms of over-determined causation, stating that ‘an event
may be causally overdetermined either because two conditions each sufficient though not necessary in the
circumstances for its occurrence are present together (additional causes) or because they are so related that if one
had not been present the other would have been (alternative causes)’ (Hart and Honoré (n7) xl). Wright’s
terminology of ‘duplication’ and ‘pre-emption’ is adopted here for two reasons. First, given the emphasis on the
importance of clear use of technical terminology, it is preferable to avoid the term ‘alternative causes’ which is
sometimes used with a different meaning in the caselaw as a contrast with ‘cumulative causes’. Secondly, Wright’s
account benefits from greater analytical clarity on this point because, as he observed, ‘Hart and Honoré also
submerged and sometimes confused the critical distinction between duplicative and pre-emptive causation by
constructing an overlapping typology of overdetermined cases’, so they did not always maintain a clear distinction
between the two forms of over-determination (Wright, ‘The NESS Account’ (n3) 287).
54
55
74
3.3.1
But-for, NESS and pre-emption
Pre-emption describes a scenario where two acts could each have been sufficient to cause the
harm, but one occurs before the other so the second one never gets to run its course. This is
illustrated by the classic desert traveller scenario: A puts poison in the water keg of a desert
traveller which is sufficient to kill him if he drinks it but before he drinks the water, B empties
the keg, and the traveller dies of thirst. The but-for test leads to the conclusion that neither act
was a cause of the traveller’s death: if he had not died of thirst he would have died from
poisoning and vice versa. Effectively, the but-for test ‘produces the absurd result that the plaintiff’s
injury was uncaused’.58
It has been suggested that the but-for test would identify B as the cause of the desert traveller’s
death if the outcome was described more precisely as ‘death when and as it happened’ i.e. death
through dehydration at that particular moment. This tactic of ‘detailing the injury’, Wright
argues, is ‘nothing more than proof by tautology’ because ‘[t]he factors believed to be causally
relevant…are incorporated into the description of the manner of occurrence of the injury…, and
they are then demonstrated to be causally relevant because we cannot construct that precise
description without them’.59 It is therefore clear that the but-for test cannot cope adequately with
pre-emptive causation.
The legal solution to this absurd outcome would be to impose liability on one or both parties
despite the failure to satisfy the legal test for causation. Such a solution is driven by the
instinctive knowledge that the death cannot be ‘uncaused’, one or both parties must be
responsible for the death despite the difficulty of establishing a causal link. Thus reliance on the
but-for test effectively turns the legal solution to pre-emption into a value judgment as to
responsibility. But, as Hamer has argued, ‘the over-determination problem is one of fact and
58
59
Beever (n41) 331.
Wright, ‘Causation’ (n2) 1778.
75
philosophy rather than value and policy’.60 This is also an unnecessarily complicated ‘solution’
because there is not really a ‘problem’ regarding the existence of a causal link and this can be
seen by explaining and applying the NESS test.
In contrast to the but-for test, the NESS test asks whether the negligence was a ‘necessary
element of a set of antecedent actual conditions that was sufficient for the occurrence of the
consequence’.61 In the desert traveller scenario, by poisoning the water flask A created the
potential for a set to occur that would be sufficient to bring about the traveller’s death. However,
for the set of conditions to be complete, in other words for it to actually be sufficient, the
traveller must drink the water containing the poison. Until he drinks the poison there is not an
actually occurring set of conditions sufficient to cause his death that contains A’s act of adding
poison to the water. Since B then emptied the water (and therefore also the poison) the traveller
never actually drank the poison. This means that A’s act was prevented from becoming part of
an actually occurring set of conditions that were together sufficient to cause the death, so A is not a
cause of the harm.62 A set of conditions jointly sufficient to cause the death now exists
containing just B’s act of emptying the flask. This set of conditions runs its course so it is an
actually occurring set, so B is a cause.
3.3.2
But-for, NESS and duplication
The plurality of causes means that one outcome could be caused by a variety of sufficient sets
and usually only one of these sets will actually have occurred, but over-determination in the form
of duplication arises when more than one sufficient set of conditions actually occurs.
Duplication, in its simplest form, is illustrated by the classic ‘double-hit hunters’ example: Hunter
A fires his gun and the bullet hits a walker. The walker dies and the bullet was sufficient to cause
David Hamer, ‘Chance would be a fine thing’: Proof of Causation and Quantum in an Unpredictable World’
(1999) 23 Melb U L Rev 557, 571.
61 Wright, ‘Causation’ (n2) 1790 (emphasis added).
62 See Wright, ‘Causation’ (n2) 1795.
60
76
his death. In the ordinary course of events, Hunter A would be considered to have caused the
death because his bullet was sufficient to result in the death. However, Hunter A’s bullet was
‘duplicated’ by Hunter B simultaneously firing his gun and also hitting the walker with a bullet
that was sufficient to cause the death.
If we apply the but-for test to determine whether Hunter A was a cause of the walker’s death
then the answer is ‘no’. But-for Hunter A’s shot, the walker would still have died because Hunter
B’s bullet was sufficient to kill him, so Hunter A made no difference to the outcome and was not
a cause of the death. Likewise, but-for Hunter B’s shot, the walker would still have been killed by
Hunter A so Hunter B made no difference to the outcome and is not a cause. Once again, the
but-for test ‘produces the absurd result that the plaintiff’s injury was uncaused’.63
In contrast, the NESS test fully incorporates the plurality of causes and asks whether the
negligence was necessary for the sufficiency of a sufficient set that actually occurred.64 It
therefore allows for the existence of more than one actually occurring set. If we subtract Hunter
B’s shot we are left with a set of conditions containing Hunter A’s shot, along with wind speed
and direction, the walker’s presence, the walker still being alive etc., that are sufficient to cause
the walker’s death, and Hunter A’s shot is necessary for this set to be sufficient. Likewise if we
subtract Hunter A’s shot we are left with a set of conditions containing Hunter B’s shot that is
sufficient to cause the walker’s death and Hunter B’s shot is necessary for the sufficiency of this
set. The plurality of causes, as identified by Mill, means that there are also many other possible
sets that would hypothetically have been sufficient to cause the walker’s death had they occurred
at that moment, for example a tree toppling onto him, but none of these other sets actually
occurred. However, the fact that two sufficient sets independently and simultaneously have
actually occurred does not prevent either set from being a cause. Usually only one set will actually
occur, but on the rare occasion that two sets actually occur then both are causes.
63
64
Beever (n41) 331.
See text to (n61).
77
Whereas the but-for test told us that Hunter A made no difference to the outcome because the walker
would have died anyway so Hunter A is not a cause, the NESS test tells us that in order for the
walker’s death to have been prevented Hunter A’s shot would have to be absent so Hunter A is a
cause, and that Hunter B’s shot would also have to be absent so Hunter B is also a cause. 65 This
is a subtle yet significant shift in emphasis and one which better reflects the reality that given the
plurality of causes it is possible, albeit rare, for two or more causes independently to occur at the
same time.
It is clear that the NESS test, rather than the but-for test, accurately reflects and tests the
qualities that were shown by Hume and Mill to characterise causes. The NESS test has therefore
been described as ‘a real contribution to legal analysis. It provides an extremely helpful way of
conceptualizing the nature of causal problems, and it offers a rational process for identifying
causes in over-determined cause cases’.66 This means that on a theoretical level it would be
preferable to adopt the NESS test. It represents a subtle shift in emphasis from the but-for test
which treats as a cause ‘something in the absence of which harm would not have come about’, to
calling a cause ‘something such that in the prevailing conditions harm comes about’ under the
NESS test.67 This may seem subtle but it is very significant because it accurately translates the
meaning of causation and the characteristics of causes into a workable test. In practice the NESS
test will frequently collapse into the but-for test because most cases involve straightforward
causal models, but NESS is to be preferred because it is able to tackle complex issues of
causation in a way that the but-for test cannot. This choice is ‘not governed by policy
considerations, but rather by how well each test corresponds with our intuitive concept of
causation’ based on the philosophical explanation of causation.68
Stapleton ‘Choosing’ (n1) 438.
David A Fischer, ‘Insufficient Causes’ (2005) 94 Ky L J 277, 281.
67 Honoré (n39) 385.
68 Wright, ‘Pruning’ (n8) 1020.
65
66
78
4. Overcoming potential problems with NESS
The final section of this chapter will use the NESS test to overcome some common legal
problems that are currently addressed using the Wardlaw test of ‘material contribution to harm’,69
but first the penultimate section addresses some potential problems with the NESS test. The
problem with the Wardlaw test is that it is not clear how exceptional it is; it may be an exception
to the but-for test that compensates for the conceptual inadequacies of that test, or it may
involve a relaxation of the causation requirement itself. This problem arises because it is unclear
precisely what causal problem is addressed by the test. The NESS test would similarly face
potential problems in more complex cases if causation were analysed in isolation from the
negligence inquiry. This section therefore uses the damage doctrine to incorporate a more
nuanced idea of ‘outcome’ into Wright’s NESS test. In the final section it will then be seen that
the NESS test allows us a level of conceptual clarity that is absent in the Wardlaw test.
Wright has explained that duplicated causation can take a more complex form than that
encapsulated in the double-hit hunters scenario. The double-hit hunters scenario involves
‘symmetrical duplication’ since each shot is part of a set of conditions that is sufficient for the
walker’s death independently of the other shot. Duplication is ‘asymmetrical’ where there are
various sources of the harmful agent and the total dose exceeds a sufficient amount (hence
causation is over-determined) but each individual source is dependent on other sources to make
a sufficient set. Since each source is not independently sufficient but is dependent on other
sources, the duplication is said to be ‘asymmetrical’. Wright explains this problem more fully
using the example of river pollution, which will be explained below. He explains that the NESS
test can be used to address asymmetrical duplication, but his application of the NESS test seems
to be so inclusive that it risks becoming useless as a functioning test for causation in negligence.
It is important, however, to apply the NESS test within the context of the negligence inquiry,
and defining the damage precisely at the outset shows that asymmetrical duplication is actually a
69
Bonnington Castings v Wardlaw [1956] AC 613 (HL).
79
very limited causal problem. In this limited problem, the NESS test is very inclusive but is also
accurate. Again it is important to remember that causation is not determinative of liability;
causation is a factual relation so the contours of the test for it cannot be driven by the perceived
fairness of the outcome. The limits of liability must be addressed by other doctrines, and here
there are issues relating to quantification and to the effects of joint and several liability.
4.1
Defining key causal terms
A distinction is drawn in negligence cases between so-called ‘divisible’ and ‘indivisible’ damage.
This terminology broadly corresponds to the scientific terminology of ‘stochastic’ or ‘nonstochastic’ disease. An indivisible, or stochastic, disease is an ‘all-or-nothing’ phenomenon:
There is almost always a clear distinction between people who have the disorder and
those who do not. The disease can differ in its severity from one case to another, but the
factors that determine its severity are in general quite different from those that determine
whether or not it occurs at all.70
In contrast, non-stochastic diseases ‘occur in a continuum of severity with no clear distinction
between cases and non-cases’ although ‘[d]octors may adopt a dichotomous case definition to
facilitate the practical management of patients’.71 In other words, the disease gradually worsens
and at a particular level of severity it is deemed to exist although ‘the chosen threshold is to some
extent arbitrary’.72 Divisible disease is therefore dose-related and each exposure to the causal
agent makes the illness more severe. For example in Thompson v Smiths Shiprepairers,73 the
claimant’s loss of hearing was divisible because it gradually worsened with exposure to noise.
Similarly the claimant’s pneumoconiosis in Cartledge v E Jopling,74 and the claimant’s asbestosis in
David Coggon and Anthony Newman Taylor, ‘Causation and attribution of disease in personal injury cases: a
scientific perspective’ [2009] JPI Law 12, 13.
71 ibid.
72 ibid.
73 Thompson v Smiths Shiprepairers (North Shields) Ltd [1984] QB 405 (QBD).
74 Cartledge v E Jopling & Sons Ltd [1963] AC 758 (HL).
70
80
Holtby v Brigham & Cowan75 were divisible diseases because they became more severe the more
the claimant inhaled silica dust or asbestos respectively.
As Coggan and Taylor explain, the term ‘stochastic’ means relating to chance, so the probability
of suffering a stochastic disease depends on exposure to risk factors. This means that although it
is not possible to state whether an individual will develop a particular disease, it is possible to say
how their probability of developing the disease is affected by particular factors. So ‘the potency
of a cause can be quantified in terms of the “relative risk” that it carries for the disease…For
example, smoking 20 cigarettes per day for 40 years might increase the risk of lung cancer 15fold in comparison with a non-smoker – a relative risk of 15’.76 In contrast, because nonstochastic diseases gradually develop in severity, we cannot sensibly talk about the probability of
developing the disease and instead ‘the potency of a cause is most meaningfully quantified by its
average impact on the severity of a disease’.77 The term ‘divisible’ is effectively a lay-term to
describe the relationship between the disease and its causes in non-stochastic diseases where
each exposure to the harmful agent makes the disease worse.
It is important to note that describing the outcome as divisible is not the same as describing the
exposure to sources or triggers of the harmful outcome as ‘divisible’ or ‘cumulative’.78 If the
exposures are ‘divisible’ in the sense that there were a number of exposures, simultaneous or
spread over a period of time, the damage may still be indivisible if it occurs once a particular
threshold has passed but does not continue to increase in severity with the increase in dose.79
Holtby v Brigham & Cowan (Hull) Ltd [2000] 3 All ER 421 (CA).
Coggon and Taylor (n70) 14.
77 ibid.
78 See also Sienkiewicz v Greif (UK) Ltd [2011] UKSC 10, [2011] 2 WLR 523, [12]-[14] (Lord Phillips)
79
This will be revisited in relation to terminology used to describe Wardlaw since some commentators use the terms
alternative and cumulative to signal that the exposures are divisible/indivisible, others use these terms to signal
divisibility of damage.
75
76
81
4.2
The problem of asymmetrical duplication
To illustrate asymmetrical duplication, Wright considers a river pollution scenario where five
units of pollution were necessary and sufficient for the injury that followed, and each of seven
defendants discharged one unit of pollution.80 He explains two details in a footnote. First, he
‘assume[s] that the injury was not accelerated or aggravated by the extra units of pollution. If it
was, causal contribution would be even clearer’.81 Second, he assumes ‘that the units of pollution
arrived simultaneously at the site of the injury. Obviously, if five units arrived before the other
two and produced the injury before the other two arrived, the first five units were causes of the
injury and the last two were not. Their potential effects were pre-empted by the effects of the
first five’.82 The NESS test, as will be explained later, considers each unit to be a cause. These
details, however, are immensely significant in understanding the causal problem so they should
not be confined to a footnote. The facts must be addressed in more detail before addressing the
application of the NESS test to them.
First of all, it is essential to identify whether the injury was aggravated by the extra units of
pollution. If each unit aggravated the injury, i.e. made it more severe, then the damage occurs
along a continuum of severity and is ‘divisible’.83 As explained in Chapter One, Stapleton insists:
It cannot be over-emphasised that the formulation of the ‘damage’ forming the gist of
the action defines the causation question. Logically one can only deal with causation after
one knows what the damage forming the gist of the action is.84
It is essential to identify the damage at the outset, and to determine whether it is divisible or
indivisible. If the damage is divisible then each unit of pollution causes a portion of the damage.
So where the damage is divisible, the duplication problem simply does not arise because each
Wright, ‘Causation’ (n2) 1793.
ibid.
82 ibid.
83 If it was ‘accelerated’ then the damage is actually still indivisible, this point will be revisited later in relation to
indivisible damage.
84 Jane Stapleton, ‘The Gist of Negligence: Part 2 The Relationship Between ‘Damage’ and Causation’ (1988) 104
LQR 389, 393.
80
81
82
unit causes the damage to be more severe.85 (This does not mean causation in divisible damage
cases is entirely unproblematic, indeed it will be analysed in more depth later, it just means that
where the damage is divisible we do not face a problem of duplication.)
It should also be noted that the problem of asymmetrical duplication, as the name implies, is
unique to duplication rather than pre-emption, as Wright commented in his footnote. If the units
of pollution were added consecutively rather than simultaneously then those units that were
added after the harm occurred (the death of the fish) would be pre-empted from becoming
causes. They could never form part of a causally sufficient set that actually occurred.
In practical terms this means that the problem of asymmetrical duplication is confined to very
limited circumstances. It only arises where the damage is indivisible and various sources of the
harmful agent combine to form an indecipherable mass so that they operate simultaneously.
4.2.1
The solution to asymmetrical duplication
In the river pollution example of asymmetrical duplication, application of the but-for test would
lead to the conclusion that none of the individual units was a cause of the injury in the sense of
strong necessity because the remaining six units would still have caused it. The but-for test once
again suggests that the injury was ‘uncaused’.
Yet the solution under the NESS test is not as simple as it is in the double-hit hunters scenario.
In that scenario, the duplication is symmetrical because each shot is independently sufficient to
cause the injury. To avoid the injury, the each shot would have to be absent. But where the
duplication is asymmetrical as in this river pollution scenario, each unit is not independently
sufficient, so we cannot say of any individual unit that it must be absent if the injury is to be
avoided. Since no individual unit is either strongly necessary (a but-for cause) or independently
sufficient (symmetrical duplication) its presence or absence seems to be irrelevant to the
outcome. Yet the same could be said of each of the seven units, returning us to a position where
Martin Hogg, ‘Developing Causal Doctrine’ in Richard Goldberg (ed) Perspectives on Causation (Hart Publishing
2011) 42.
85
83
the harm seems to be mysteriously ‘uncaused’, and logically we know this is not true. However,
Wright explains that this is not because of a weakness in the NESS test, it is because the NESS
test is being misapplied. If the NESS test is applied correctly, each unit is still a cause of the
injury:
[E]ach defendant's one unit was necessary for the sufficiency of a set of actual antecedent
conditions that included only four of the other units, and the sufficiency of this particular
set of actual antecedent conditions was not affected by the existence of two additional
duplicative units.86
This is because of the plurality of causes whereby there can be more than one sufficient set for
any outcome, and it is possible for more than one set actually to occur. As noted in an earlier
section, the philosophical account of causation involves strong sufficiency (or its equivalent,
weak necessity) so it does not require independent sufficiency. Instead of asking whether a
particular unit would have to be absent for the harm to be avoided (strong necessity), the correct
question is whether the set containing that unit would have to be absent for the harm to be
avoided (weak necessity). The situation appears complicated. However, this is only because the
two sets are not independently sufficient so there is an overlap between sets.
Wright continues to explain that this conclusion is logically the same whether each of the seven
units was contributed by a different individual, or whether one person contributed five units and
another contributed two units.87 Since the units combine in an indecipherable mass the causal
status of a unit of pollution is not dependent on whether the remaining units were contributed
by one person or by many people. The two units contributed by the second person are still
‘necessary for the sufficiency of a set of actual antecedent conditions that included only three of
the first defendant’s five units, a set whose sufficiency was not affected by the existence of two
Wright, ‘Causation’ (n2) 1793.
Richard Wright, ‘Once More into the Bramble Bush: Duty, Causal Contribution, and the Extent of Legal
Responsibility’ (2001) 54 Vand L Rev 1071, 1106.
86
87
84
additional duplicative units also provided by the first defendant’.88 The fact that the two sources
occur at the same time, or combine to form an indecipherable whole, makes this disaggregation
of the five unit contribution into its component units logically possible: if we are able notionally
to disaggregate the defendant’s contribution from the other contribution then logically we can
also notionally disaggregate the other contribution.
Wright thus argues that the same causal situation exists whether the harmful agent is one that
would normally be thought of as existing in ‘units’ or not, so if two or more fires merge to
destroy a building, each is a cause of the destruction even if it would have been insufficient on its
own, and even if one of the other fires alone would have been sufficient. The second fire ‘was
necessary for the sufficiency of a set of actual antecedent conditions which included another
fire…that was “at least large enough to be sufficient for the injury if it merged with a fire the size
of the second fire”’.89 It will be argued below that in principle Wright is correct to assert that
disaggregation is appropriate even when the harmful agent is something such as fire that would
not normally be thought of as existing in units. However the practical example that he uses to
support this assertion is inappropriate because the harm in cases of fire may be divisible so it
should not be used to support an argument concerning an indivisible harm.90
Wright has since come to question this ‘at least enough’ formulation, describing it as ‘overly
demanding’.91 This description is surprising given that his ‘at least enough’ formulation seems to
make the NESS test very inclusive because in cases of duplication it allows any contribution,
which is neither strongly necessary nor independently sufficient, to be labelled as a cause by
constructing a set containing that contribution and just enough of the other contributions. He
explains:
Wright, ‘Causation’ (n2) 1793.
ibid 1793.
90 For example if two fires combine to damage a house it may be the case that one fire would have damaged one
room, the second fire another room so the damage would be divisible, but if a person was trapped and killed by the
fire their death would be an indivisible damage.
91 Wright, ‘The NESS Account’ (n3) 291.
88
89
85
My initial elaborations of the NESS account were overly demanding. I incorporated the
weak-necessity requirement in the definition of singular instances of causation. As I have
previously stated, this it too restrictive. The weak-necessity requirement is sufficiently
incorporated in a properly formulated causal law, which contains in its antecedent only
those abstract conditions the instantiation of which is necessary for the sufficiency of the
set of conditions that is sufficient for the immediate instantiation of its consequent.
When analysing singular instances of causation, an actual condition c was a cause of an
actual condition e if and only if c was a part of (rather than being necessary for) the
instantiation of one of the abstract conditions in the completely instantiated antecedent
of a causal law, the consequent of which was instantiated by e immediately after the
complete instantiation of its antecedent, or (as is more often the case) if c is connected to
e through a sequence of such instantiations of causal laws. This formulation of the
requirement for a NESS condition is more straightforward and simpler to apply than my
initial formulations, which requires ‘at least so much’ description of actual conditions in
some situations in order to (validly) treat other conditions as NESS conditions.92
This is unconvincing. This seems like a looser approach to causation where the NESS test allows
us to understand whether a condition is capable of being a cause but does not allow us to
determine whether it actually was a cause on a particular occasion. The NESS ‘test’ thus
conceived seems to explain causal generalisations but not to actually test whether something was
a cause in a particular instance. This would mean that it has little or no value as a legal test where
the law requires a causal link to exist between the particular defendant’s negligence and the
particular claimant’s loss.
However, Wright’s previous ‘at least enough’ formulation is appropriate in cases of asymmetrical
duplication because asymmetrical duplication only arises where the sources of the harmful agent
are factually combined into an indecipherable mass. Since the units are physically combined
92
ibid.
86
together their effects overlap so we are able to construct overlapping sufficient sets of units
containing A’s unit and ‘just enough’ of the other units, or B’s unit and ‘just enough’ of the other
units. It is impossible to say that five particular units had an effect and the remaining units had
no effect because they were all present together and all operated at the same time – they formed
an indecipherable whole so the effects of each unit are similarly indecipherable.
The ‘at least enough’ formulation would be unnecessary and inappropriate in other scenarios that
do not involve indivisible damage and sources of the harmful agent combining in an
indecipherable mass. If only five units of pollution are added to the water, one by A and four by
B then each unit was necessary for the sufficiency of the set, so A’s unit was a cause of the harm
occurring. This is a simple scenario like Barnett because only one sufficient set has actually
occurred so there is no duplication. If A added 5 units and B simultaneously added 5 units then
the scenario is equivalent to the double-hit hunters case of symmetrical duplication so both A
and B would be labelled as causes on the NESS test because they are independently sufficient. If
A added 5 units and B later added 5 units then A’s units may pre-empt B’s units from being
causes so there is no duplication. Likewise if a total of seven units were added one after the other
then the first five units may pre-empt the final two units from becoming causes. There is no
danger of the ‘at least enough’ formulation being so expansive that it included cases of preemption because the NESS test correctly requires the unit to be necessary for a sufficient set of
conditions that actually occurred. One ingredient in any sufficient set must be the condition that
‘the damage has not already occurred’. It is impossible to construct a sufficient set containing
some of A’s units and some of B’s in a pre-emptive situation like this, because as soon as A’s 5
units have caused the damage we lose the requisite ‘the damage has not already occurred’
condition in relation to any of B’s units.
The NESS test is very inclusive here, for example if one litre of pollution was sufficient to cause
the injury and A added a teaspoon (five millilitres) of pollution while other parties added
pollution up to a level of several litres, then the NESS test still leads us to the conclusion that
87
A’s teaspoon of pollution was a cause of the damage. Yet it is important to remember that this
problem of asymmetrical duplication only arises when a) the harm is indivisible and b) the
defendant contributes a source of a harmful agent that combines with other sources to form an
indecipherable mass so that they operate simultaneously. It is also important to remember that
causation is only one element of legal liability. Although liability would be joint and several, the
defendant can seek contribution from the other responsible tortfeasors. Additionally, a
defendant is not liable in negligence if his causal contribution was de minimis, so this reflects the
idea that interpersonal responsibility is narrower than causal responsibility so a defendant will not
be liable merely because he was a cause of the harm.
This section has addressed asymmetrical duplication which, along with symmetrical duplication,
only arises in cases of indivisible damage. The next section looks at divisible damage to highlight
the potential problems that can arise with the NESS test.
4.2.2
Causation of divisible damage
Divisible damage can be explained through a variation on the river pollution example. This time,
instead of saying that an indivisible injury occurs at a level of five units of pollution, divisible
disease is the equivalent of a situation where each unit of pollution will kill, for example, ten fish.
If A adds one unit of pollution and B adds three units of pollution, then in total there will be
four units of pollution so forty fish will die. The death of forty fish could be the outcome to
which we apply the NESS test. In order for the set of factors to be sufficient to kill forty fish
there must be four units of pollution so each unit was necessary for the sufficiency of the set.
This would mean that A and B were both a cause of the outcome.
This does not seem to be the best way to understand the outcome. A corrective justice-based
system of negligence liability is focused on interactions between individuals. Even though a
claimant might be seeking compensation for his total loss, if this total loss consists of a number
of smaller losses each of which is attributable to a different individual wrongdoer, then there
88
have been a number of individual interactions. An individual wrongdoer is only responsible for
loss that he caused, and has no moral responsibility for entirely separate loss caused by a separate
wrongdoer. So in the pollution example of divisible damage A’s one unit constituted a set that
caused ten fish to die, and B’s three units constituted a separate set that was sufficient to cause
thirty fish to die. The presence of A’s unit and B’s units are in no way related to each other, they
have occurred independently, so their respective harms should also be treated independently.
This is clearer in the following scenario: A shoots a walker and hits his arm, B shoots the walker
and hits his leg. The shots were independent and the harms are independent. A caused the loss
of the arm, B caused the loss of the leg. These examples seem potentially different because in the
shooting scenario the causes remain separate and the effects are visibly separate too. In contrast,
in the river scenario the causes cannot be separated once in the water and the effects are not
visibly different. Although this means that A’s unit of pollution cannot be tied to ten specifically
identified fish, this is a difficulty of proof and does not change the underlying causal model
which is one where A has only caused the death of ten fish. This is important because in
corrective justice-based interpersonal responsibility a person can only be morally responsible for
an outcome if he was a cause of that outcome, so it would be wrong to impose moral
responsibility on A for fish deaths that were caused by B.
In cases involving divisible harm the courts achieve corrective justice by ‘apportioning’ liability
based on causal responsibility:
In tort law, the cause of action only arises upon the occurrence of “damage”: the legal
forensic process is in fact best truly looked at in the reverse sequence…by defining at the
outset the “damage” it is claimed has been caused by the fault of the defendant. In cases
of divisible damage, legal responsibility for each part of the damage can then be traced
back using the tortious mechanisms of causation, blameworthiness and duty.93
Simon Brown and Susan Lindsey, ‘Seeing the wood for the trees – Loftus-Brigham and apportionment of damage’
(2005) 21 Const LJ 431, 434.
93
89
But it important to be clear that this is not an exception to joint and several liability, rather it is a
recognition that each defendant caused a specific portion of the total loss so each is liable only
for that loss that he caused. In Holtby,94 the claimant developed asbestosis after having been
exposed to asbestos over the course of his employment with different employers. He had
worked for the defendant employer for approximately half of his working life. Asbestosis is a
divisible disease which means that the defendant only caused a portion of the disease. Although
they were not able to assess this contribution precisely, the court made a twenty-five percent
reduction in damages, stating ‘[t]he court must do the best it can to achieve justice, not only to
the claimant but the defendant, and among defendants’.95 This is the correct approach to have
taken, but the justification ought to have been made more forcefully. Apportionment achieves
justice not simply because it seems unfair to make a single defendant shoulder the burden when
other employers also caused the claimant’s illness. Corrective justice demands that the defendant
only pay for the loss that he has caused, and where the disease is divisible he has only caused a
portion of the overall illness. As Hogg has argued, this means that a request from the defendant
for apportionment should not be necessary but the ‘apportionment’ should take place as a matter
of course.96He says that ‘in many cases this will have to be a “best guess” approach, but to refuse
to attempt such an apportionment is to saddle a sole defender with damages for losses which he
did not cause’.97
The precise contribution made by the defendant in Holtby could not be established, but justice
was better achieved by taking a broad brush approach to attempt apportionment than if there
had been no apportionment at all. This approach was followed in Allen v British Rail Engineering
Ltd (BREL),98 where the claimant suffered ‘vibration white finger’ from the use of vibrating tools
during his employment. The tools were used throughout his employment but this was only
Holtby (n75).
ibid [20] (Stuart-Smith LJ).
96 Martin Hogg, ‘Causation and apportionment of damages in cases of divisible injury’ (2008) 12 Edin LR 99, 101.
97 ibid 104 (emphasis added).
98 [2001] EWCA Civ. 242, [2001] ICR 942.
94
95
90
negligent on the part of his employer once the risk of vibration white finger became known. The
illness is one which worsens with the longer vibrating tools are used so it is divisible and
apportionment was made even though the precise contribution of the negligence could not be
established. Once again, corrective justice was more closely achieved by attempting imprecise
apportionment than by denying apportionment completely.
4.2.2.1 Divisible damage and interaction between causes
The aetiology of divisible diseases is not always straightforward because there is not always a
linear correlation between the potency of the cause and the severity of the disease. This is known
as ‘effect modification’ but this is not an insurmountable challenge. Effect modification describes
the phenomenon whereby a particular causal factor becomes more potent when combined with
another causal factor and affects both divisible and indivisible diseases.99 For example if smoking
carries X relative risk of lung cancer, and exposure to asbestos carries Y relative risk of lung
cancer, the relative risk of lung cancer in a smoker who has been exposed to asbestos is not
necessarily simply X+Y, but may be higher because the interaction between smoking and
asbestos may be an ‘effect modifier’. Similarly, in non-stochastic (divisible) diseases where the
disease occurs along a continuum of severity, repeated exposure to one harmful agent, or
exposure to that agent in combination with another harmful agent may increase the severity
exponentially rather than incrementally. To return to the example of river pollution, say each unit
of pollutant A would kill 5 fish, and each unit of pollutant B would kill 5 fish, but a combination
of a unit each of pollutant A and B would kill 15 fish. The harm can still be divided into 5 dead
fish for which A’s pollutant forms a NESS set, 5 dead fish for which B’s pollutant forms a NESS
set, and 5 more dead fish for which the NESS set comprises both A and B’s pollutants.
Coggan and Taylor explain that the task of apportionment remains ‘relatively straightforward’
even in cases involving effect modification. This can be illustrated by considering the
99
Coggon and Taylor (n70) 14.
91
apportionment that the court achieved in the case of Rahman v Arearose Ltd.100 In this case the
court undertook a detailed analysis of the various harms suffered by the claimant and determined
which were causally attributable to the negligence of the first defendant alone, the second
defendant alone and to both defendants’ negligence together. The claimant was assaulted at work
by two black youths who injured his right eye and it was held that the first defendant had
negligently failed to protect him against this attack. The claimant then received negligent medical
treatment from the second defendant which led to the permanent loss of sight in his right eye. It
is clear that each defendant was causally responsible for distinct injuries, but the claimant also
subsequently suffered various psychiatric illnesses and the court needed to apportion liability for
these illnesses. The expert evidence showed that the claimant suffered a phobia of black people
which was caused by the assault alone, a depressive disorder which was a reaction to the loss of
the eye, post-traumatic stress disorder attributable to the assault and the operation, and enduring
personality change which was due to the synergistic effect of the both causes. The court was thus
able to apportion liability for each head of damages based on causal responsibility. It is clear that
courts are able to undertake the task of apportionment in cases of divisible harm even where the
various causes interact with a synergistic effect. This ought to be done in all cases of divisible
harm in order to achieve corrective justice in these cases.
5. Using NESS to overcome common problems with exceptional legal tests
In negligence law the but-for test is supplemented with the Wardlaw test of ‘material contribution
to harm’,101 so it is often said that the claimant must prove that the defendant’s negligence
‘caused or materially contributed to’ the damage.102 It is unclear in the caselaw and academic
literature whether the Wardlaw test is an application of the but-for test or an exception to it. If it
is an exception to the but-for test it is further unclear whether it is an exception to the factual
[2001] QB 351 (CA).
Wardlaw (n69).
102 ibid 620. Note that this phrase pre-dates the decision in Wardlaw but this is the leading case on the test of
material contribution to harm.
100
101
92
causation requirement, or whether it is a test of factual causation that compensates for the
conceptual inadequacy of the but-for test. This has led Honoré to criticise it as an ‘indefinite, if
not indeterminate’ notion which is ‘purely pragmatic and leaves the theoretical problem
untouched’.103 This uncertainty, concerning just how exceptional the Wardlaw test is, has been
important in the development of the law since the Wardlaw test was used as a kind of stepping
stone between the but-for test and the McGhee/Fairchild test of ‘material contribution to the risk
of harm’.104 The uncertainty is most significant because it is unclear what causal problem is
addressed by the test so its scope of application and its effects are ill-defined.105
It will be argued in this section that the Wardlaw test is sometimes an application of the but-for
test and sometimes an exception to it. Where it departs from the but-for test this is because of
the conceptual inadequacies of the but-for test, so rather than being an exception to the
causation requirement the Wardlaw test actually ensures that factual causation is tested more
accurately than is allowed by the but-for test. This can be illustrated by drawing on the analysis
of the NESS test that has been developed throughout this chapter and in section 4 in particular.
Since the NESS test better reflects the underlying philosophical account of the concept of
causation, it allows the lacunae in the but-for test to be pinpointed. In straightforward cases the
but-for test works but only because it is shorthand for the NESS test, so it needs to be
supplemented in more complex cases. It will be suggested that it is preferable simply to use the
NESS test in those cases rather than to use the Wardlaw test which can only be understood
properly by drawing on the NESS analysis and which risks adding to confusion by introducing
its own vocabulary surrounding ‘contribution’.
Honoré (n39) 364.
The McGhee/Fairchild test is the subject of Chapter Four.
105 It is worth noting that this is not identical to the American concept of a ‘substantial factor’ since that seems to
apply only to cases of symmetrical duplication (see David W Robertson, ‘Causation in the Restatement (Third) of Torts:
Three Arguable Mistakes (2009) 33 Wake Forest L Rev 1007, 1018). This means that the ‘substantial factor’ solution
addresses the double-hit hunters problem of duplicative causation but it is not clear that it would apply in a case
such as Wardlaw which involved a build up of the harmful substance in unknown quantities. The scope of the
‘substantial factor’ test is unclear because it has not been scrutinised in technical detail, instead ‘difficult
puzzles…have been simply handed over to the jury’ (Stephen Bailey, ‘Causation in negligence: what is a material
contribution?’ (2010) 30 LS 167, 169). Caution must therefore be exercised in referring to American scholarship on
the practical implementation of the NESS test.
103
104
93
It is first necessary to present the facts and decision in Wardlaw, and to identify the specific areas
of confusion surrounding its application and effects.
5.1
Bonnington Castings v Wardlaw
The claimant in this case was exposed to silica dust in the air at work coming from two sources.
The majority of the dust, the ‘innocent’ dust, was produced by the hammer operated by the
claimant and it was not possible to fit a dust extraction plant there. This was therefore a nonnegligent source of silica dust in that it was unavoidable. The remainder, the ‘guilty’ dust, was
present due to the negligence of the defendant in failing to provide adequate extraction at the
swing grinders in the workshop. The claimant inhaled this silica dust in the air at work and
developed pneumoconiosis, a disease of the lungs.
As Bailey has explained, the decision of the House of Lords cannot be fully understood without
presenting the previous decisions and the point of appeal.106 At first instance the Lord Ordinary
had found that failure to provide respirators was a cause of the disease, which suggests that all
the dust was considered ‘guilty’ at first instance. On appeal, the defendant argued that only the
dust from the swing grinders was a result of negligence, and that the claimant had failed to
establish causation in respect of this source of dust since the dust from the innocent source was
far greater in quantity so was the ‘most probable’ cause of the disease. The Court of Session did
not reach the question of whether the negligence concerned only the dust from the swing
grinders, and disposed of the case on the point of causation. They held that exceptionally the
onus of proof lay on the defendant to prove that the dust from the swing grinders could not
have been a cause of the disease, and the defendant had failed to prove this. The issues to be
decided in the House of Lords were therefore first whether the Court of Session had been
correct to reverse the onus of proof, and second, if the onus of proof lay on the defendant,
whether he could discharge this by proving that the innocent dust was the most probable cause
because it was the far greater source of dust.
106
Stephen Bailey, ‘Causation in negligence: what is a material contribution?’ (2010) 30 LS 167
94
The House of Lords held that the onus of proof was not reversed, it remained the task of the
claimant to prove that the negligence had caused or materially contributed to the damage. On the
facts, both sources had ‘materially contributed’ to the injury. The guilty dust had made a
‘material’, that is more than de minimis, contribution to the illness.
Lord Reid explained that the medical evidence was that pneumoconiosis is caused by ‘a gradual
accumulation in the lungs of minute particles of silica inhaled over a period of years’. 107 He
continued:
That means, I think, that the disease is caused by the whole of the noxious material
inhaled and, if that material comes from two sources, it cannot be wholly attributed to
material from one source or the other. I am in agreement with much of the Lord
President's opinion in this case, but I cannot agree that the question is: which was the
most probable source of the respondent's disease, the dust from the pneumatic hammers
or the dust from the swing grinders? It appears to me that the source of his disease was
the dust from both sources, and the real question is whether the dust from the swing
grinders materially contributed to the disease. What is a material contribution must be a
question of degree. A contribution which comes within the exception de minimis non
curat lex is not material, but I think that any contribution which does not fall within that
exception must be material. I do not see how there can be something too large to come
within the de minimis principle but yet too small to be material.108
Lord Keith noted also that the defendant’s argument was based on the lack of evidence to show
the precise proportions of dust coming from each source.109
Wardlaw (n69) 621 (Lord Reid).
ibid 621 (Lord Reid).
109 ibid 626,
107
108
95
5.1.1
The scope of the test
Part of the problem with this decision seems to be that the significance that it has acquired
extends beyond the specific points of appeal that it actually addressed. The objective of the
appeal was to determine whether the onus of proof was reversed, and whether the defendant
could prove the absence of a causal link by showing that the innocent source was the ‘most
probable’ cause. It was the reversing of the onus of proof by the Court of Session that was the
unusual and unorthodox approach. By contrast, there was nothing controversial about holding
that the claimant should prove that the negligence ‘caused or materially contributed to’ the
disease. The claimant argued that ‘inhalation of the dust was the cause of the disease. If [the
defendants] were shown negligently to have contributed a part of the dust which was not
negligible, [the claimant’s] case was proved’.110 As Bailey has observed, ‘[n]o authorities were
cited for this proposition and there was no suggestion that it was regarded as novel’.111 This
means that little attention was paid to the exact boundary between proof that the negligence
‘caused’ and proof that it ‘materially contributed to’ the disease, so the definitions and scope of
application have been gleaned through subsequent analysis of the case and its application in later
decisions. This means that factual aspects of the case, such as the ‘cumulative’ exposure to dust,
and the divisibility of the disease, the lack of precise evidence as to the proportions of ‘guilty’ and
‘innocent’ dust, as well the absence of apportionment in the case, have been relied upon as
determining the scope of the test of material contribution to harm.
5.1.1.1 Divisibility of damage
It is unclear what significance the divisibility of damage has for the applicability of the test of
material contribution to harm. The court did not determine whether the claimant’s disease was
divisible. It was called a ‘disease of gradual incidence’ which suggests it is divisible, but Lord
Keith also said that without the negligent dust the claimant ‘would not have developed
110
111
ibid 617.
Bailey (n106) 172.
96
pneumoconiosis when he did and might not have developed it at all’,112 which suggests that it is
indivisible.
Pneumoconiosis was later held to be divisible,113 so Hogg has suggested that Wardlaw ‘settled the
point that, for a defender to have caused a pursuer’s injury, it is not necessary that the defender
was the sole cause of the injury, but merely that, but-for the defender’s conduct, the injury would
not have occurred to the same extent’.114 If the fact that pneumoconiosis is divisible implies that
the ‘material contribution to harm test’ only applies to divisible injuries then it was a mistake for
the court not to have apportioned damages in Wardlaw.
Yet it has also been noted that where damage is divisible and the defendant has only caused a
portion of the total loss there should be an apportionment of damages. The absence of
apportionment in Wardlaw has led some to suggest that the disease was treated as though it were
indivisible, so the divisibility of the disease is not the reason for the application of the test. Thus
Smith LJ (extra-judicially) has said ‘material contribution to harm’ was a modification of the butfor test.115 In her view, the disease in Wardlaw was treated as indivisible but today it would be
treated as divisible and apportioned. When the damages are apportioned it shows that the
negligence was a but-for cause of a portion of loss. The absence of apportionment in Wardlaw
could be taken as an indication that the but-for test was not satisfied in that case.116
Bailey notes, however, that damages had been assessed at first instance on the basis that
provision of respirators would have prevented the claimant inhaling dust from both sources, so
if damages were to be apportioned later the case would have to have been remitted to the Court
of Session to consider evidence as to the appropriate measure of damages. It may have been the
case that the state of the evidence would not have permitted such an assessment, and it
Wardlaw (n69) 626.
Nicholson v Atlas Steel Foundry & Engineering Co Ltd [1957] 1 WLR 613.
114 Hogg (n96) 99.
115 Janet Smith, ‘Causation – the search for principle’ [2009] JPI Law 101, 101.
116 ibid 102.
112
113
97
‘presumably would not have been worth it in terms of cost’.117 This means that the case ‘cannot
be read as holding that…the harm must as a matter of law be regarded as indivisible’.118 It has,
however, been applied in cases of divisible and indivisible damage, for example the claimant in
Bailey suffered an indivisible disease, brain damage.119 In this case, however, the causal process
was still a ‘cumulative’ one in that the brain damage was a result of the claimant’s weakened state
and this weakness had been caused by a number of factors. The next section therefore turns to
consider the relevance of the causal process for the test of material contribution to harm.
5.1.1.2 ‘Cumulative’ causal process
In Wardlaw, Lord Reid observed that the disease was caused by ‘the whole of the noxious
material inhaled’.120 Lord Keith also noted that the negligent exposure was continuous over a
long period of time and that ‘it was the atmosphere inhaled by the pursuer that caused his illness
and it is impossible…to resolve the components of that atmosphere into particles caused by the
fault of the defenders and particles not caused by the fault of the defenders, as if they were
separate and independent factors in his illness’.121 This has been taken as suggesting that the test
of material contribution to harm applies where the exposure to the harmful agent is ‘cumulative’
rather than ‘alternative’. These terms do not have an agreed meaning though.
Green suggests that the Wardlaw test applies to ‘cumulative causes’ but she equates ‘cumulative
causes’ with ‘divisible injury’ so she takes the view that cumulative causes can be independent
(such as in Performance Cars122) or interdependent as in Bailey.123 The claimant in Bailey was in a
weakened state which, when she vomited, caused her to be unable to respond naturally to the
vomit, instead aspirating it leading to cardiac arrest and brain damage. The defendant hospital
Bailey (n106) 173.
ibid.
119 Bailey v Ministry of Defence and another [2008] EWCA Civ. 883, [2009] 1 WLR 1052
120 Wardlaw (n69) 621.
121 ibid 626.
122 See Chapter One (n137)
123 Sarah Green, ‘Contributing to the risk of confusion? Causation in the Court of Appeal’ (2009) 125 LQR 44.
117
118
98
had negligently failed to resuscitate her adequately following an operation, thus causing her to be
weaker than she would otherwise have been. At the same time she developed pancreatitis which
also caused her to become weaker but this was a natural complication and not attributable to the
negligence of the hospital. The question was whether the negligent failure to resuscitate the
patient after her first operation was a cause of her ultimate injury. Green argues that Bailey
effectively involved divisible harm because ‘weakness was undoubtedly divisible, since patients
are not classed simply as being either weak or strong, but lie somewhere along an almost
infinitely calibrated spectrum’.124 However, in Bailey, the weakness was the cause of the harm
which was the claimant’s eventual brain damage due to aspiration of her vomit. The damage here
is indivisible, even though its cause – weakness – came from two sources with each contributing
to the degree of weakness. Others regard the causal process as forming the distinction between
divisible and indivisible injury, for example Gullifer has suggested that harm is divisible if there is
a temporal or spatial separation between the causes as in Performance Cars, and indivisible if the
candidate causes are cumulative causes of the same type that combine together as in Wardlaw.125
Stauch says that where the disease is indivisible it should not matter whether the causal process is
cumulative or alternative.126
It is impossible to determine the scope of the test from the facts of Wardlaw alone because the
limited grounds of appeal mean that the above issues simply were not addressed.
5.1.2
Analysis of the Wardlaw test
Bailey seeks to show that the test of material contribution to harm is an application of the butfor test. It is important that he regards the but-for test as being synonymous with factual
causation, so any departure would be exceptional. For him, the but-for test must be satisfied, and
when we say that a claimant must prove that the negligence ‘caused or materially contributed to’
ibid 45.
Louise Gullifer, ‘One cause after another’ (2001) 117 LQR 403, 406.
126 Marc Stauch, ‘“Material contribution” as a response to causal uncertainty: time for a rethink’ (2009) 68 CLJ 27,
29.
124
125
99
the loss, this is a more elaborate way of saying that the claimant must prove but-for causation
that resolves ‘the need to find words that cover both cases arising (effectively) from single causes
and those arising from multiple causes’.127 We know, however, that an outcome does not have a
single cause. This means that when Bailey says that ‘caused’ relates to cases arising effectively from
single causes he is making recourse to the Hart and Honoré type approach that uses common
sense to distinguish ‘causes’ from ‘mere conditions’. He is selecting between all of the necessary
conditions, so his approach is evaluative rather than purely factual. This is why he finds it
significant to remind us that in any given causal set, more than one factor may be identified as ‘a
cause’ rather than as a ‘mere condition’.
The NESS analysis in sections four and five showed that there are a range of causal problems
that the but-for test is unable to resolve because of its conceptual inadequacies. Causation, as
with any aspect of negligence, can also give rise to evidential problems, and in Wardlaw Lord
Keith noted that the defendant’s case rested partly on the fact that it was not possible to prove
the precise proportions of silica dust attributable to each source. The defendant had also argued
that the greater source of dust was the ‘most probable’ cause, but this confuses the question of
what the claimant must prove with the standard to which he must prove it i.e. on the balance of
probabilities.128 It is therefore also possible that the test of material contribution to harm seeks to
reinforce this distinction and reiterate that an outcome is caused by a set of conditions and each
condition is equally deserving of the label ‘cause’ as Mill showed in his work on the philosophy
of causation.
The meaning of the Wardlaw test of ‘material contribution to harm’ will necessarily be obscured
if it responds both to conceptual inadequacies of the but-for test and to evidential difficulties.
The preferable solution is therefore to adopt the NESS test since it is conceptually more sound
than the but-for test and able to resolve causal problems that the but-for test cannot. This would
mean that evidential issues arising from the balance of probabilities standard of proof, or from
127
128
Bailey (n106) 178.
See also Bailey (n119) [56] (Waller LJ).
100
weaknesses in the evidence available to the court, could be readily identified for what they are.
The final section of this chapter presents a practical example of a recent decision where the
Wardlaw test was applied and seeks to illustrate the advantages of using the NESS test.
5.2
Practical application of the NESS test
The NESS analysis of causation enables the causal question to be clearly articulated, making the
resolution of cases more straightforward. This final section considers a number of cases where
there has been confusion in application of the Wardlaw test and in the question of
‘apportionment’ and seeks to show that approaching these cases via the NESS test clarifies the
issues.
5.2.1
The question of causation in Bailey
The decision of the Court of Appeal in Bailey was notable for showing confusion over the
meaning and applicability of the Wardlaw test of material contribution to harm. Waller LJ, with
whom Sedley and Smith LJJ concurred, approached the issue by asking ‘was this a case in which
the judge was entitled to depart from the but-for test?’129 As this chapter has argued, the legal
issues cannot be fully articulated if the but-for test is taken as the frame of reference, because the
but-for test is conceptually inadequate. Using the NESS test makes the resolution of the case
simpler and reduces the potential for future confusion.
As noted above, the claimant in Bailey was in a weakened state and therefore aspirated her vomit
leading to cardiac arrest and brain damage. The defendant’s negligence had led to weakness, and
she was also weakened by a naturally occurring illness. The question was whether the negligent
failure to resuscitate the patient after her first operation was a cause of her ultimate injury.
The first step is to determine whether the damage is divisible; the damage in Bailey was the brain
injury which in this case was indivisible. The cause of the brain damage was the claimant’s
129
Bailey (n119) [36] (Waller LJ).
101
weakened state which prevented her from responding naturally to her vomit. Applying the NESS
test, the question is whether the weakness that was due to the defendant’s negligence was a
necessary element of a sufficient set that actually occurred? Since the patient gradually became
weakened and this overall weakness led to her failing to respond to her vomit, it is clear that the
weakness formed an ‘indecipherable mass’ because the effects operated on her simultaneously. It
was the overall weakened state that caused the damage, and the negligence contributed to the
total level of weakness, so it was a cause of the claimant’s brain damage. The judge at first
instance seems to have understood that this was a relatively simple issue, explaining that
although he could not say ‘whether the contribution made by [the negligence] was more or less
than that made by the pancreatitis…the natural inference is that each contributed materially to
the overall weakness and it was the overall weakness that caused the aspiration’, so he found that
the causal link had been established.130
In the Court of Appeal, Waller LJ eventually concluded that causation could be established using
the test of material contribution to harm, but his reasoning was more complex since he
considered this test to be an exception to the but-for test. He therefore asked whether it was a
case in which he was entitled to depart from the but-for test which led him to weigh up the
policy arguments and to seek to distinguish the case from a range of other exceptional
approaches such as the loss of chance argument and the Fairchild exception. Waller LJ saw the
following issue as central:
It is important to be clear precisely what Wardlaw decided. Did it decide that in a
cumulative cause case where the inadequacies of medical science meant that the relative
potency could not be established all a claimant had to establish was a ‘material’
contribution?...or did a claimant still have to establish that ‘but for’ the contribution of
the negligent cause, the injury would not have occurred?131
130
131
Bailey v Ministry of Defences, Portsmouth Hospitals NHS Trust [2007] EWHC 2913 (QB), [61] (Foskett J).
Bailey (n119) [39] (Waller LJ)
102
This phrasing of the issue has a number of weaknesses: it assumes that proof of a material
contribution does not satisfy the but-for test; it does not address the nature of the damage; and it
assumes that the Wardlaw approach was necessitated because it involved cumulative causes and
because the relative contributions could not be measured with precision. Waller LJ did not
engage with these questions in any depth, but was satisfied that Lord Rodger had established in
Fairchild that ‘in cumulative cause cases such as Wardlaw the but-for test is modified’.132 As
discussed in this chapter, however, the NESS analysis shows that where damage is divisible
‘material contribution to harm’ translates as applying the but-for test (as shorthand for the NESS
test) to a portion of the overall damage, and that where the damage is indivisible it also translates
as applying the but-for test where the victim was only exposed to the threshold amount of the
harmful agent. It is only where indivisible damage occurred and the causes are duplicative that a
test of material contribution would be an exception to the but-for test. But in these cases, if the
Wardlaw test is applied consistently with the NESS test, then it makes up for the conceptual
inadequacy of the but-for test and is not an exception to the factual causation requirement. This
means that there is no need to consider, as Waller LJ did, whether a departure from the but-for
test is justified on policy grounds. Waller LJ considered arguments that have been raised in the
medical negligence context, but that relate to claims based on the loss of chance idea or on the
Fairchild exception, and these arguments are exceptional. In other words, by failing to observe
that the but-for test is conceptually inadequate and needs to be supplemented in order to
accurately test factual causation, Waller LJ made life difficult for himself and weighed up policy
factors that extend beyond corrective justice and which are unnecessary.
5.2.2
The question of ‘apportionment’
Another problem arising out of the uncertainty surrounding the reasons for the application of
the Wardlaw test is that it has led to uncertainty whether ‘apportionment’ of damages is
132
Ibid [43] referring to Fairchild v Glenhaven Funeral Services [2002] UKHL 22, [2003] 1 AC 32, [129] (Lord Rodger)
103
appropriate. If the courts apply the test of material contribution because of a lack of evidence as
to the precise proportions of the harmful agent attributable to each source then, it has been
argued, it would be inappropriate for the court to apportion liability because it would lack a
rational basis on which to calculate the portion of the damage that is attributable to the
defendant’s negligence. As already discussed, damages should be ‘apportioned’ where the injury
is divisible. Since each source of the harmful agent causes a portion of the total loss, this is a
false apportionment – it is not an apportionment of liability but a question of quantifying that
portion of the loss attributable to the defendant’s negligence. It is therefore the definition of the
damage, rather than the NESS test, that determines the ‘apportionment’ issue. The NESS test,
however, is still valuable because it highlights that the Wardlaw test is needed to respond to a
conceptual weakness in the but-for test and not because of an evidential problem.
In Hatton v Sutherland,133 a case involving psychiatric illness resulting from occupational stress, the
claimant established causation by showing that the defendant’s negligence made a material
contribution to her illness and in the Court of Appeal Hale LJ concluded that where there were
multiple causes of a psychiatric injury then apportionment should be attempted. In the decision
of the Court of Appeal in Dickins v O2 plc,134 however, it was suggested that where the test of
‘material contribution to harm’ is applied it would be inappropriate to apportion liability
regardless of whether the harm was divisible or indivisible. The reason for this conclusion was
the court’s understanding that the ‘material contribution to harm’ test is used where the various
contributions made by the defendant(s) cannot be quantified.
[I]n a case which has had to be decided on the basis that the tort has made a material
contribution to harm but it is not scientifically possible to say how much that
contribution is (apart from the assessment that it was more than de minimis) and where
133
134
[2002] EWCA Civ 76, [2002] 2 All ER 1.
[2008] EWCA Civ 1144, [2009] IRLR 58.
104
the injury to which it has lead is indivisible, it will be inappropriate simply to apportion
the damages across the board.135
Since the extent of the contribution cannot be quantified, the court reasoned that any
apportionment would necessarily be artificial:
That means that a claimant can succeed on causation even though he cannot
demonstrate what the causative potency of the tort was, save to say that it had some
effect beyond the minimal. It seems to me that, if in one breath the judge holds that all
that can be said about the effect of the tort is that it made an unspecified material
contribution, it is illogical for him, in the next breath, to attempt to assess the percentage
effect of the tort as a basis for apportionment of the whole of the damages.136
It is not at all clear that the reason for applying the Wardlaw test is that the relative contributions
cannot be quantified. As this chapter has illustrated, the Wardlaw test may also represent an
application of the but-for test to a portion of the total loss in cases of divisible damage, and may
make up for the conceptual inadequacies of the but-for test in over-determined cases of
indivisible damage. Since the nature of the problem addressed by the Wardlaw test is unclear it is
natural that there will be confusion in its application. As this chapter has shown, the first step in
any case is to define the damage as being divisible or indivisible. Once this has been done, the
NESS test can be applied to either type of damage and can resolve problems of pre-emption and
duplication. If the damage is indivisible then the negligence is a cause of the whole of the loss so
there is no apportionment vis-à-vis the claimant, and joint and several liability applies. If the
damage is divisible and the negligence has caused only a portion of the total loss then
apportionment is required in principle because the defendant is only causally responsible for a
portion of the loss so only that portion of the loss is ‘wrongful’ in corrective justice. If there is an
insurmountable evidential barrier to the apportionment exercise then this must be addressed
squarely.
135
136
ibid [46] (Smith LJ).
ibid [43] (Smith LJ).
105
Chapter One showed that causation has a vital but limited role in interpersonal responsibility and
this chapter has drawn this issue out in greater detail to isolate factual causation, the fact of being
a cause, from evaluative conclusions that a condition was the responsible cause. It was shown
that the NESS test is preferable to the but-for test as a test of factual causation because it is
better matched to the philosophical account of what it means to be a cause and is therefore able
to resolve more complex causal problems that the but-for test is unable to resolve. Adopting the
NESS test would ensure that courts are equipped to address a wider range of causal scenarios
without resorting to exceptional tests. This is important because it has shown that cases such as
Wardlaw and Bailey may involve an exception to the but-for test but they do not involve an
exception to the causation requirement. Having addressed the conceptual aspects of causation in
this chapter the remaining chapters turn to evidential problems that have arisen and consider
how they can be resolved consistently with principles of corrective justice. The work in this
chapter will be valuable because the NESS analysis means that the evidential problems can be
more clearly identified. The final chapters will also be strengthened by being able to draw on the
analysis of the role that causation has within interpersonal responsibility and of the roles of the
other negligence doctrines in addressing the remaining aspects of interpersonal responsibility.
106
Chapter Three: Loss of a chance
Introduction
The loss of a chance argument has been raised by claimants who have been unable to prove on
the balance of probabilities that the defendant’s negligence was a cause of their physical damage.
The claims in Hotson v East Berkshire Health Authority,1 and Gregg v Scott,2 arose in the medical
negligence context and in both cases the negligence consisted of the doctor’s failure to diagnose
and treat an existing illness. In both cases the court found that on the balance of probabilities the
patient’s existing illness could not have been cured with careful treatment, so the delay was not a
cause of any eventual physical damage. The claimants argued instead that although it was less
than probable that their condition could have been cured, the ‘chance’ of a cure had been
reduced by the negligence and this ‘chance’ was something of value. They therefore sought to
reformulate the actionable damage as the ‘chance’ of avoiding physical injury rather than the
physical harm itself. If it succeeded, this action for loss of a chance would lead to proportionate
recovery of damages for the physical harm, for example in Hotson the loss of a 25 percent chance
of avoiding avascular necrosis was valued at 25 percent of the total loss caused by the avascular
necrosis. Although this argument therefore centres on the question of whether a ‘chance’ of a
better outcome can constitute actionable damage in negligence, it is an attempt to sidestep the
difficulties of proof of causation of the physical harm on the balance of probabilities standard of
proof.
It will be argued in this chapter that the House of Lords was right to reject the claim for
proportionate recovery for the loss of a chance of a better medical outcome in both of these
cases. In corrective justice-based interpersonal responsibility, liability is for ‘wrongful loss’ and
the loss can only be characterised as wrongful if it was, in fact, caused by the defendant’s
Hotson v Fitzgerald and others [1985] 1 WLR 1036 (QBD); Hotson v East Berkshire Health Authority [1987] 2 WLR 287
(CA); Hotson v East Berkshire Health Authority [1987] AC 750 (HL).
2 Gregg v Scott [2002] EWCA Civ 1471, [2003] Lloyd's Rep. Med. 105 (CA); [2005] UKHL 2, [2005] 2 AC 176 (HL).
1
107
negligence. Causation is a factual relationship, so the negligence either was or was not a cause of
the harm. The claimant need only persuade the court of this fact on the balance of probabilities
and if the court is, on balance, persuaded that the negligence was a cause of the damage then it
accepts the fact of causation to be proved and the claimant recovers in full. Although the loss of
chance argument ostensibly reformulates the damage as the chance of avoiding physical harm, it
will be argued that what it in fact achieves is to ‘discount’ liability to reflect the degree of doubt
over the fact of causation.
However, it will also be argued in section 4.2 that the doctor/patient relationship raises unique
issues of interpersonal responsibility because the patient’s ‘chance’ is valued by both parties and
has intrinsic value independently of the physical outcome. At the time of diagnosis and
treatment, the outcome of treatment is uncertain and beyond the control of the parties – the
doctor and patient both regard him as having a ‘chance’ in these circumstances and they do what
they can to improve his chance – the doctor gives him careful treatment, indeed the purpose of
his duty of care when a patient is already ill is to take an uncertain situation and do what he can
but he cannot guarantee cure. Both parties therefore take the statistical medical assessment of the
likelihood of cure as being something that is valuable to the claimant so a reduction in the
statistical chance of cure is a loss of something of value and ought to be recognised as actionable
damage. Importantly it is valued for its own intrinsic worth rather than as a proportion of the
physical harm. While the practical effect is that loss of chance would ‘assist’ claimants by
providing compensation even where the doctor’s negligence is not proved to have caused
physical harm, this compensation would be very modest and is certainly not a way to bypass the
causation requirement regarding the physical harm.
1. Analysis of Hotson v East Berkshire Health Authority
The purpose of the following section is to set out the facts of Hotson and to clarify some of the
issues that arose in that case. The claim for the loss of a chance of avoiding injury was successful
108
at first instance and in the Court of Appeal but was ultimately unsuccessful in the House of
Lords. A later section of the chapter will seek to evaluate the loss of a chance argument. The
current section takes the preliminary step of setting the facts of the case within the framework of
the conventional understanding of negligence. The purpose is to set aside any possible
misconceptions about how the traditional principles of causation operate. This includes building
on Chapter One to separate clearly the concepts of damage, causation and quantification. This
will allow an analysis of the loss of chance argument to be based on a clear understanding of the
concepts involved and of precisely where the problem lies. Three main strands of the case
emerge here: the definition of damage and its relation to the so-called ‘hook’ argument; the
relationship between damage and quantification, particularly in relation to the vicissitudes
principle; and the different role played by damage in contract and tort. The relationship between
a statistical assessment of the probability of causation and the balance of probabilities standard
of proof will be considered in a later section in relation to both Hotson and Gregg.
1.1
The facts
The claimant in Hotson was a boy who had fallen from a tree and sustained an injury to his hip,
specifically a fracture to the left femoral epiphysis. He was taken to hospital (for which the
defendant health authority was responsible) where he was examined, his knee was x-rayed and
bandaged. The medical staff negligently failed to x-ray his hip so they did not diagnose the injury
to his hip and instead sent him home untreated. He continued to suffer severe pain for the next
five days after which he returned to hospital where his hip was x-rayed, the correct diagnosis was
made, and emergency treatment begun. The boy suffered avascular necrosis in the epiphysis
which involved deformity of the hip, with loss of movement and a limp, and would worsen in
future due to osteoarthritis developing in the joint because of the injury. The femoral epiphysis is
a layer of cartilage between the head and neck of the femur and avascular necrosis of the
epiphysis develops because of a failure of the blood supply to the epiphysis. The award of
109
damages for the five day period of suffering was unproblematic. The difficulty arose in relation
to the avascular necrosis. The injury clearly constituted actionable damage, the defendant owed
the claimant a duty of care and had breached that duty. The difficulty was in establishing on the
balance of probabilities that the negligence was a cause of the injury. At first instance, Simon
Brown J found that there was only a 25 percent chance that the claimant would have avoided
avascular necrosis if his injury had been treated appropriately, so a conventional claim would
have failed. However, he awarded the claimant damages to compensate the loss of a 25 percent
chance of avoiding avascular necrosis. These damages were 25 percent of the value of the overall
injury. He effectively treated the case as involving an issue of quantification of the loss caused by
the delay in diagnosis, leading to proportional recovery, rather than as involving a difficulty of
establishing a causal link between the delayed diagnosis and the final injury. The Court of Appeal
upheld the claim, but it was rejected by the House of Lords. The House of Lords did not reject
the concept of the loss of a chance, but held that in this instance the finding of fact was that, on
the balance of probabilities, at the time of the negligence insufficient blood vessels remained
intact for the injury to be treatable. Thus he was destined to suffer avascular necrosis and there
was no ‘chance’ that had been lost.
With regard to the causal process involved in this case, the judge at first instance had found that
avascular necrosis occurs when there are insufficient blood vessels to keep the epiphysis alive.
When the claimant first went to hospital some blood vessels remained intact although distorted,
but the delay in commencing treatment meant that already ruptured blood vessels continued to
bleed into the joint thus increasing pressure in the joint and blocking the intact blood vessels.
The delay in diagnosing and treating the injury therefore caused damage to those blood vessels
that had remained and sealed the claimant’s fate by turning his injury into an inevitability.3
The judge thought it ‘possible but improbable’ that sufficient vessels remained intact after the fall
because the injury to the joint was at the upper end of the spectrum of severity. He found that
3
Hotson v Fitzgerald and others [1985] 1 WLR 1036 (QBD), 1040 (Simon Brown J).
110
there was a 25 percent chance that following the fall sufficient blood vessels had remained intact
to keep the epiphysis alive and avoid avascular necrosis. Lord Mackay, in the House of Lords,
suggested that Simon Brown J had arrived at this figure because it was midway between the
figures given by the two expert witnesses.4 Simon Brown J said himself that he was ‘unattracted
to, and finally unable to accept, either of the competing extreme views’,5 so his conclusion does
seem to achieve a compromise between the views of the two expert witnesses.
1.2
Distinguishing damage, causation, and quantification
As argued in Chapter One, it is essential to maintain a clear distinction between the negligence
doctrines, notably damage, causation, and quantification. At first instance, Simon Brown J said
that ‘[i]n the end the problem comes down to one of classification. Is this on true analysis a case
where the plaintiff is concerned to establish causative negligence or is it rather a case where the
real question is the proper quantum of damage? Clearly the case hovers near the border’. 6 The
conventional approach to negligence is that a claimant must show that he has suffered actionable
damage and that the negligence was a cause of this damage. Physical injury is a recognised form
of damage, but the loss of a chance of avoiding physical injury was previously not recognised as
being capable of forming the gist of a negligence action. Once all the elements of a negligence
claim have been established and the court is concerned with valuing the claimant’s loss, the
principle of restitutio in integrum means that the claimant should be compensated for all the losses
that flow from the personal injury. Certain lost chances are recoverable at this stage. Simon
Brown J explained:
Time and time again courts evaluate past and future medical risks and award damages
based on an assessment of the likelihood (a) of some adverse medical condition, like
epilepsy or osteo-arthritis, developing consequent on the injury, or (b) that some preHotson v East Berkshire HA [1987] AC 750 (HL).
Hotson v Fitzgerald and others [1985] 3 All ER 167, 173 (Simon Brown J’s evaluation of the evidence is available in
this report but omitted from that of the Weekly Law Reports).
6 Hotson (QBD) (n3) 1043.
4
5
111
existing, perhaps degenerative, condition would in any event have manifested itself so as
to cause the same or at any rate some lesser degree of disability as has been occasioned
by the injury.7
This is simply an aspect of returning the claimant to his pre-tort position. But the judge then
went on to overstate the significance of this, suggesting that:
There is really no significant difference between that exercise and what the court is being
invited by the plaintiff to do in the instant case.8
In fact, there is a significant difference between these principles and the loss of a chance
argument in that they address issues of quantification whereas the loss of chance argument
involves a prior issue of damage and causation. The following sections will therefore distinguish
the aspects of damage, causation and quantification in those traditional principles in order to
show that the loss of a chance of avoiding avascular necrosis is not analogous with existing
negligence principles.
1.2.1
The ‘hook’ argument
Simon Brown J considered that the claimant’s five days pain and suffering constituted actionable
damage, and that the lost chance of recovery could simply be regarded as a question of
quantification.9 This is referred to by Stapleton as the ‘hook’ argument, the idea being that once
it has been proved that the negligence caused some physical damage, albeit minimal, any lost
chance of avoiding further harm can be ‘hooked’ on to this claim as an issue of quantification.10
But he went on to say ‘this very point underlines how unsatisfactory it would be to suppose that
the case should turn entirely on whether there is any directly provable injury, however slight.
ibid 1045.
ibid.
9 Hotson (QBD) (n3) 1045.
10 Jane Stapleton, ‘Cause in fact and the scope of liability for consequences’ (2003) 119 LQR 388; ‘Loss of the
chance of cure from cancer’ (2005) 68 MLR 996.
7
8
112
That will itself often be a matter of chance’.11 However Simon Brown J has overstated the ‘hook’
concept here. In his earlier explanation of conventional principles he had noted that courts
regularly assess the likelihood that the claimant will develop a medical condition consequent on the
injury.12 It is necessary to show that there is a causal link between the injury that the defendant
has actually caused and the anticipated future illness. The defendant is not required to pay for all
future illnesses that the claimant might suffer, he is required to compensate for those illnesses
that the claimant might suffer because of the harm the defendant has negligently caused him. In
Hotson the claimant was unable to prove that the delay in treatment had caused the avascular
necrosis, so he would have been equally unable to prove that his five days pain and suffering was
causally related to the avascular necrosis. Lord Bridge explained this in the House of Lords:
The damages referable to the plaintiff’s pain during the five days by which treatment was
delayed in consequence of failure to diagnose the injury correctly, although sufficient to
establish the authority’s liability for the tort of negligence, have no relevance to their
liability in respect of the avascular necrosis. There was no causal connection between the
plaintiff’s physical pain and the development of the necrosis. If the injury had been
painless, the plaintiff would have to establish the necessary causal link between the
necrosis and the authority’s breach of duty in order to succeed. It makes no difference
that the five days’ pain gave him a cause of action in respect of an unrelated element of
damage.13
By understanding that the hook argument would only succeed where there is a causal
relationship between the actionable injury and the risk of future illness it is clear that it is not
arbitrary as Simon Brown J suggested. The hook idea should continue to be used where
actionable injury causes the claimant an increased risk of future illness, but cannot apply in
Hotson (QBD) (n3) 1045-46.
See text to (n7) (emphasis added).
13 Hotson (HL) (n4) 780 (Lord Bridge).
11
12
113
Hotson where there is no causal relationship between the five days pain and suffering and the
increased risk of avascular necrosis.
1.2.2
The vicissitudes principle
As noted above, Simon Brown J also observed that courts regularly take into account the
likelihood that some pre-existing condition would have resulted in the same degree of disability
when valuing the loss caused by the defendant.14 This is known as the vicissitudes principle.
When calculating loss of earnings the multiplier used by the court is reduced to take account of
the ‘vicissitudes of life’ i.e. the risk that the claimant’s earnings would have been reduced by
other unrelated accidents or illnesses.15 Simon Brown J said that this amounts to holding that ‘if
the risk, or chance [of avoiding injury], is less than 50 per cent then the plaintiff gets nothing. If,
however, it is over 50 per cent then the court should proceed to determine the matter as one of
quantum, which involves having regard to the chances and contingencies and making discount
accordingly’.16 This, he said, ‘smacks somewhat of heads I win, tails you lose’.17
He is right to say that such a position would discriminate in favour of defendants, yet once again
he has overstated the issue. The discount made for the vicissitudes of life is made to reflect as
accurately as possible the true value of the loss suffered by the claimant. Damages should return
the claimant to the position he would have been in without the negligence so the court must do
its best to account for the events that would have befallen him anyway including the likelihood
that unrelated illness or accident would have impacted on the claimant’s earning capacity in the
future. It is an issue of quantification. The fact that the defendant’s negligence was a cause of the
injury actually suffered is not in doubt. In Hotson the claimant did not establish that the
defendant’s negligence caused him to suffer avascular necrosis so the subsidiary task, of taking
See text to (n7)
Lim Poh Choo v Camden & Islington AHA [1980] AC 174 (HL)
16 Hotson (QBD) (n3) 1049.
17 ibid 1049.
14
15
114
the pre-existing chance of injury into account when valuing the loss caused by the negligence,
never arises.
The difference is apparent if Hotson is compared with Smith v Leech Brain.18 In that case the
defendant’s negligence caused the claimant to suffer a burn on his lip which triggered a precancerous condition and led to his eventual death. The pre-cancerous condition meant that
independently of the negligence there was an increased likelihood that he would have developed
cancer in the future, so his damages were reduced by five percent to account for this possibility.
Stapleton suggests that this is indistinguishable in effect from the loss of chance-based claim for
proportionate recovery:
The “discount” is made not to reflect that chance that the triggering had been due to a
cause other than the defendant’s fault (because it was clear that this had not been the
case) but to reflect the true value of the loss which the defendant had caused the plaintiff
to suffer. In many cases, if not all, this approach will give a result indistinguishable from
that produced had the claim been framed in terms of loss of a chance.19
Even though the result might look indistinguishable from a claim for proportionate recovery
framed as a claim for the loss of a chance, it does not mean, as Stapleton suggests, that valuing a
lost chance as a proportion of the physical harm to which the chance relates is conceptually
sound. There was no doubt in Smith that the defendant’s negligence had caused the claimant to
suffer the harm at that particular time. Yet when making a discount for the vicissitudes of life at
the valuation stage, account was taken of the pre-cancerous condition to personalise the
likelihood that the claimant would have died prematurely anyway. In Hotson the doubt
surrounded the question of whether the defendant’s negligence had caused him to suffer
avascular necrosis at this time. If proportional recovery were allowed in Hotson it would not
reflect the chance that the claimant would have suffered the same loss due to an unrelated illness
Smith v Leech Brain & Co [1962] 2 QB 405 (QBD).
Jane Stapleton, ‘The gist of negligence: Part 2 the relationship between “damage” and causation’ (1988) 104 LQR
389, 398.
18
19
115
at some stage in the future, it would reflect the degree of doubt over whether the defendant had
even caused him any loss.
In the House of Lords, Lord Ackner correctly stated that ‘[o]nce liability is established, on the
balance of probabilities, the loss which the plaintiff has sustained is payable in full. It is not
discounted by reducing his claim by the extent to which he has failed to prove his case with 100
per cent certainty’.20 Understanding that the vicissitudes principle relates to valuation of loss once
a causal link has been established makes it clear that it is distinct from the loss of chance
argument which seeks to solve a difficulty relating to proof of causation. Simon Brown J
suggested that the rejection of the loss of chance argument would be inconsistent with the
conventional approach in this respect, yet it has been shown that this is not the case.
Stapleton has criticised the decision of the House of Lords for failing to resolve the essence of
the claimant’s argument, ‘namely whether reformulation of the gist in terms of loss of a chance
should now be acceptable’.21 It is clear that this is what the claimant was trying to do, and that he
was trying to do this because of his inability to prove a causal link between the negligence and
the physical outcome. The above discussion should also have added some clarity to what
happens at each stage of a conventional claim in terms of damage, causation and valuation, so
that arguments for and against reformulating the gist in terms of loss of a chance are considered
against an accurate picture of traditional recovery.
1.3
The contract/tort distinction
At first instance Simon Brown J noted that loss of chance is actionable in contract law but
rejected the contract/tort distinction as a basis on which to reject loss of chance. In his view, the
particular branch of the law of obligations in which the claim is brought is not relevant to
weighing the merits of the case since ‘[t]his would be entirely artificial, as is well illustrated by
20
21
Hotson (HL) (n4) 793 (Lord Ackner).
Stapleton, ‘The gist of negligence’ (n19) 393.
116
recognising that it would turn, for instance, on whether the patient was consulting his doctor
privately or under the National Health Service.22
In principle a coherent distinction can be drawn between the position in contract and in tort
because the doctrine of damage has a different role in each of these branches of law. For a claim
to succeed in the tort of negligence the claimant must bring together the elements of actionable
damage, duty of care, breach of duty, factual and legal causation. This means that he must prove
that he has suffered a legally recognised form of loss in order to have a complete claim. Only
once he establishes all the elements of the tort does the question of valuing the claim arise. In
contrast, in contract law a claim becomes actionable upon breach of contract which means that
loss suffered by the claimant is relevant only to quantification and is not a necessary element for
him to have a successful claim in the first place.
Simon Brown J’s rejection of the contract/tort distinction is specific to the medical negligence
context since the same doctor/patient relationship can arise in a contractual or non-contractual
relationship. In this particular context, allowing recovery for loss of a chance in contract but not
in tort would lead to disparity in the treatment of claims depending on whether the patient was
treated by the National Health Service or a private healthcare provider. While it may be the case
that a patient can choose to pay for private healthcare and thereby purchase a higher quality of
care, it would seem nonsensical to suggest that the actual losses that a patient suffers are
different depending on whether they had a contract or not. As Reece has argued, ‘[t]he private
patient may have paid for a higher standard of care, but the standard of proof should surely
remain the same’.23
The real question is whether the ‘chance’ that a patient has of recovering from illness or injury
when he visits his doctor is something that is capable of constituting a loss. The starting point
for analysis must be the concept of chance rather than the type of legal action. So whilst it is
conceptually correct to allow for a different approach in tort and in contract because of the
22
23
Hotson (QBD) (n3) 1044.
Helen Reece, ‘Losses of Chances in the Law’ (1996) 59 MLR 188, 191.
117
different role played by damage within the different types of claim, the important question is
whether the ‘chance’ of recovering from illness is a meaningful concept, something of which the
patient can be deprived and in respect of which he ought to be able to claim against his doctor.
2. Analysis of Gregg v Scott
The loss of chance argument for proportionate recovery was subsequently raised again in the
medical negligence context in the case of Gregg v Scott.24 The claim once again failed. The
claimant in Gregg went to the defendant doctor complaining of a lump under his arm which the
defendant negligently diagnosed as a benign fatty lump rather than referring the patient for
further tests which would have shown that it was cancerous. This resulted in a nine month delay
in the diagnosis and commencement of treatment of the claimant’s disease. The claimant argued
that the nine month delay due to the defendant’s negligence had caused him to lose a chance of
being cured because his chance of cure had been reduced from a statistical 42 percent at the
time of negligence to 25 percent at the time of trial.25 The medical definition of cure is ten years
disease-free survival, so the claim was that the negligence had caused him to lose a chance of ten
years disease-free survival.26 (Since the details of the facts are relatively dense they will be
presented in more depth in relation to each issue rather than being repeated here.)
It will be argued that he presented a claim that was destined to fail. There are barriers to
accepting the loss of a chance argument in itself, but even more so in a case such as this where
the ‘chance’ has not been definitively lost. This is not because the claimant’s ‘chance’ was
reduced to 20-30 percent,27 but because the claimant was still alive, and until he either died or
Gregg v Scott [2002] EWCA Civ 1471, [2003] Lloyd's Rep Med 105; [2005] UKHL 2, [2005] 2 AC 176.
The reliability of these statistics will be considered below in addition to asking whether the loss of such a chance
can constitute damage.
26 The start date for measuring the ten year period is unclear. The medical evidence referred to ten years disease free
survival which suggests it is measured from the date that the tumour has been eliminated. Lord Phillips thought that
it probably started at the date of commencement of treatment (Gregg v Scott [2005] UKHL 2, [2005] 2 AC 176, [132]).
At trial and in the Court of Appeal reference was made to the chance of five years’ disease-free survival but this
seems to be because of the number of years that had elapsed between the negligence and the trial, so the medical
definition of cure was still premised on ten years’ disease-free survival.
27 The difficulty of giving an accurate figure will be addressed later.
24
25
118
survived disease-free for ten years it could not be said that this ‘chance’ had run its course. The
problem seems to arise because the claimant used the loss of a chance argument to try to avoid
the difficulty of proving a causal link between the negligence and the harm. But given that the
claimant had not yet suffered the ‘outcome’ to which the chance related i.e. dying before ten
years had elapsed, even if the evidence had favoured finding a causal link between the doctor’s
negligence and the claimant’s probable death, he would still have been unable to claim for this
because he had not yet suffered the loss. Effectively, faced with a case where it was going to be
difficult to prove that the defendant’s negligence had caused loss to the claimant, he decided to
advance a case based on loss of a chance, yet he made his loss of a chance argument more
difficult/less palatable than it might otherwise have been by claiming for the loss of the chance
of avoiding a harm that he had not yet, and may never, suffer. However, it seems that the
claimant could have advanced a more modest claim based on traditional principles and that such
a claim ought to have succeeded. Similarly to the discussion of Hotson in the previous section, the
following analysis seeks to clear away any misconceptions surrounding how conventional
negligence principles would apply to this case in order to be left with a clearer understanding of
what problems remain that might be solved by taking a loss of chance approach to recovery.
2.1
Pain and suffering
During the nine months between the defendant’s negligence and the eventual diagnosis of the
claimant’s illness the cancer had ‘gradually enlarged’.28 Following the referral for a biopsy he was
admitted to hospital with ‘acute and intense chest pain which was a result of the lymphoma
having spread, in particular to the left pectoral region’.29 The spread of the cancer is a personal
28
29
Gregg v Scott [2002] EWCA Civ 1471, [2003] Lloyd's Rep Med 105, [1] (Latham LJ).
ibid [2].
119
injury, and the claimant ought to have been able to recover for this physical injury and the
consequential pain and suffering.30
The next issue is the extent of the treatment that the claimant had to undergo. The claimant
underwent chemotherapy, supplemented by radiotherapy. The tumour responded but not
completely, so the claimant underwent a more aggressive treatment of high dose chemotherapy,
with harvesting of stem cells. He suffered side effects from the treatment, and these were severe
in the case of the high dose chemotherapy. He had to give up work, felt ill and had felt weak
ever since. At first instance, Inglis J found that ‘it is possible to say on the basis of Professor
Goldstone’s model that he would more probably than not have achieved complete remission
with initial CHOP therapy and without high dose chemotherapy with stem cell harvesting’. 31
Baroness Hale, despite rejecting the claim for loss of a chance, still acknowledged that this
amounted to a finding that ‘the claimant would have achieved initial remission had he been
treated earlier’.32 She ultimately considered that it was probable that the claimant would still have
suffered the same setbacks and that the relapses would have happened anyway, but she did say:
Even on conventional principles, this does not necessarily mean that the claimant is not
entitled to anything at all. The defendant is liable for any extra pain, suffering, loss of
amenity, financial loss and loss of expectation of life which may have resulted from the
delay. If, without the delay, the claimant would have achieved a longer gap before more
radical treatment became necessary, then he should be entitled to damages to reflect the
acceleration in his suffering. If the pain and suffering he would have suffered anyway was
made worse by the anguish of knowing that his disease could have been detected earlier,
then he should be compensated for that.33
Gregg (HL) (n26) Lord Hope, who ultimately would have allowed the claim for loss of a chance, recognised that he
was in the minority and that the claim would therefore fail, but expressed his hope ‘that it was not too late for the
pain and suffering which the appellant suffered due to the tumour’s enlargement…to be brought into account by
way of an award of general damages’ [123].
31 [38] (Inglis J), cited in Gregg (HL) (n26) [202] (Baroness Hale).
32 Gregg (HL) (n26) [203].
33 Gregg (HL) (n26) [206].
30
120
It is therefore clear that even if she thought that the delay had no effect on the overall prognosis
for cure, so the spread of the cancer did not affect the final outcome of death/survival, she did
think that the spread of the cancer was physical harm and that this had affected the treatment
necessary which caused him greater pain and suffering. Again, he could have recovered for this if
he had formulated his claim in terms of the spread of the cancer and the consequences of the
spread.
2.2
Loss of life expectancy
The claimant argued that the defendant’s negligence caused him to lose a chance of a cure by
reducing this chance from 42% to 25%. Given that the medical definition of cure is ten years’
disease-free survival, the claim for lost years was that the negligence had caused him the loss of a
chance of ten years’ disease-free survival. By incorporating the medical definition of cure into the
claim for lost years the claimant seems to have made it easier for the court to reject his claim.
This is because introducing a cut-off date of ten years adds a binary quality to the inquiry – he
either will or will not live for ten years, and the delay in diagnosis either has or has not prevented
him for living for ten years. If he had advanced a more modest yet conventional claim for loss of
life expectancy, which is open-ended in nature, then it seems likely that he would have been able
to establish that the delay in diagnosis did reduce his life expectancy. Admittedly his life
expectancy was already quite short at the time of the negligence, but the spread of the cancer due
to the delay reduced it even further.
Lord Phillips highlighted this problem which was created by the way the claim for lost years was
formulated:
The conventional way of determining the effect of the injury on expectation of life is as
follows. The court determines what the claimant’s expectation of life would have been
but for the injury. The court then determines what the claimant’s expectation of life is
121
having regard to the effect of the injury. The difference between the two constitutes the
“lost years”.34
It is important to note that a conventional claim for loss of life expectancy is not the same as the
‘hook’ argument. The hook argument consists of arguing that once a claimant can point to some
actionable damage that was caused by the defendant’s negligence, all other chances are
recoverable as a matter of quantification. As discussed above in relation to Hotson, a claimant will
actually only be compensated where the reduction in the chance of avoiding future illness or of
making future gain is as a consequence of the actionable damage i.e. causally related to the
actionable damage.35 This is an area where the importance of distinguishing clearly between
issues of damage, causation, and quantification, is very apparent yet often overlooked. Mr Gregg
was unable to show that there was a causal link between the spread of the cancer and the final
outcome of being cured (or failing to survive for ten years) because on the balance of
probabilities he would not have survived ten years even with timely diagnosis. But if he had
advanced a conventional claim for the physical harm constituted by the spread of the cancer, the
additional pain and suffering caused by the spread, and the more aggressive treatment, then it
would seem likely that having more extensive cancer would have reduced his life expectancy
somewhat, even if his life expectancy was already short because he was suffering from cancer at
the time of the negligence. If a claimant with an identical disease was shot and killed by a
defendant, in valuing the lost years the court would assess his life expectancy with cancer at the
time he was shot and compensate the lost years. So it is not making an impossible demand on
the court to ask them to assess his life expectancy at the time of the negligence and his life
expectancy nine months later when the cancer had spread.
Lord Phillips summarised the various ‘adverse events’ that Mr Gregg had suffered ‘beyond the
initial development of his cancer’ as: i) the spread of the cancer to the pectoral region
accompanied by acute pain; ii) high dose chemotherapy with harvesting of stem cells; iii) relapse
34
35
Gregg (HL) (n26) [131].
See text to (n13).
122
when a tumour developed in the right axilla accompanied by chemotherapy; iv) psychiatric
distress on being told that the relapse meant that he would die; v) further suspected relapse with
additional chemotherapy.36 He correctly explained that ‘the chance that the delay in commencing
his treatment has caused each of these adverse events is not the same’.37 This highlights the need
for claimants to be careful and specific in how they detail the loss for which they are claiming,
and also the need for a claimant to be realistic about what he can prove. Given that corrective
justice requires that a defendant only pay for that loss that he caused through his negligence, a
claimant will only recover if he can prove the causal link between the negligence and the damage.
It is surely preferable for a claimant to receive modest compensation for that portion of the loss
the defendant can be proved to have caused, rather than to attempt to combine the losses into
one claim for the loss of a chance of a better outcome. Lord Phillips concluded that
On balance of probability I suspect that one is now in a position to conclude that the
delay in commencing Mr Gregg’s treatment has not affected his prospect of being a
survivor but has caused him all the other adverse events which I have set out above.38
If Mr Gregg had advanced a conventional claim for these limited heads of damage it is therefore
likely that he would have been successful.
3. Statistics, chance, and the balance of probabilities standard of proof
This chapter has illustrated that it is essential to have a clear understanding of the concepts of
damage, causation, and quantification and their interrelationship before it is possible to analyse
the loss of chance argument for proportional recovery. In light of this it is clear that if courts
recognised that the loss of a chance of a better medical outcome could constitute the gist of a
negligence action, this would represent a significant departure from traditional principles.
Gregg (HL) (n26) [167].
Gregg (HL) (n26) [168].
38 Gregg (HL) (n26) [169].
36
37
123
The loss of chance argument has been raised by claimants because of difficulties they faced
proving the causal link between the defendant’s negligence and their loss. In both Hotson and
Gregg, the claimant was already unwell and the defendant’s negligence consisted of a failure to
diagnose and treat that existing illness. Proof of causation is naturally complicated in this
scenario because there is routinely a candidate cause other than the defendant’s negligence – the
court must determine whether the illness alone would have caused the harm even with careful
diagnosis and treatment. Since the civil standard of proof is the balance of probabilities, and
recovery in negligence law is ‘all-or-nothing’, it is often said that where the patient’s pre-tort
chance of a cure exceeded fifty percent the courts regard him as having been destined to be
cured, but if his chance of cure was below fifty percent the courts regard him as being doomed
to suffer the harmful outcome even with careful treatment so the doctor’s negligence is deemed
not to be a cause and the claimant receives nothing.39 In this section it will be argued that this is
based on a flawed understanding of the balance of probabilities standard of proof, and on an
overly simplistic reading of the statistical information that informs the assessment of the
‘chance’. The balance of probabilities relates to the degree of belief the judge must form in the
fact of causation,40 and statistical evidence is just one element in informing a rational belief. The
statistical evidence should be personalised to the individual as far as possible, and in order to
form rational belief based on such evidence the judge must also be persuaded of the reliability of
the evidence. The judge’s degree of belief is not easily quantified as a percentage because it
involves a qualitative assessment of the available evidence, and it would be wrong to think that
statistics are somehow more useful or objective because they express probabilities.
Understanding that the balance of probabilities relates to the judge’s degree of belief highlights
See Allan Beever, ‘Gregg v Scott and loss of a chance’ (2005) UQLJ 201, 201. See also Stapleton ‘Loss of the chance
of cure’ (n10) 997; Sarah Green, ‘Coherence of medical negligence cases: a game of doctors and purses’ (2006) 14
Med LR 1, 3.
40 See Richard Wright, ‘Proving Causation: Probability versus Belief’ in Richard Goldberg (ed) Perspectives on Causation
(Hart Publishing 2011) 205-212.
39
124
the error of proportionate recovery, which would actually equate to discounting damages to
reflect the judge’s doubt as to causation.
3.1
What must the claimant prove? What is the problem of proof?
Lord Nicholls’ willingness in Gregg to accept the loss of chance argument seems to reflect a clear
desire to help a claimant overcome evidentiary difficulties in the medical negligence context. He
explained that ‘in cases of medical negligence assessment of a patient’s loss may be hampered, to
greater or lesser extent, by one crucial fact being unknown and unknowable: how the particular
patient would have responded to proper treatment at the right time’.41 That, however, is the
nature of a counterfactual inquiry, and it is addressed by using the balance of probabilities
standard of proof. It is true that the existing disease makes it harder for a court to determine
what would have happened to the claimant. In other contexts a claimant could have expected to
carry on unharmed unless something unusual occurs, such as a defendant’s negligence; in cases
such as these where the claimant was already unwell then he would have expected to suffer harm
unless the doctor could cure his illness. The loss of chance argument is therefore a solution to
help claimants overcome the evidentiary difficulty, as Lord Nicholls makes plain:
In suitable cases courts are prepared to adapt their process so as to leap an evidentiary
gap when overall fairness plainly so requires. Fairchild v Glenhaven Funeral Services Ltd is a
recent illustration of this in a different context. In the present context use of statistics for
the purposes of evaluating a lost chance makes good sense.42
But although proof of causation is more complicated in these cases than in many straightforward
scenarios, it is not afflicted with the same kind of ‘evidentiary gap’ as existed in Fairchild. Proof of
causation was impossible in Fairchild. By contrast, in Hotson and in Gregg it was merely difficult.
These are not cases where the only available evidence was statistical; there was evidence that
could be personalised to the individual claimants. This section will explain the conventional
41
42
Gregg (HL) (n26) [27].
Gregg (HL) (n26) [31].
125
approach to what the claimant should prove, and the standard to which he should prove it, and
apply this to the evidence available in the cases to show that it was appropriate to insist on
traditional principles.
3.1.1 Damage
As argued in the previous chapters, the role of the causation inquiry in negligence is limited and
is defined by the doctrines of damage and breach of duty. It is especially important to define the
damage that forms the gist of the action from the outset because the negligence inquiry is
concerned to determine whether the defendant’s negligence was a cause of this damage, rather
than with determining more generally what would have happened absent the negligence.
This is particularly relevant to the analysis of Gregg because the outcome of the claimant’s
treatment was still prospective so, apart from the limited heads of damage noted earlier, he had
not suffered the harmful outcome to which the ‘chance’ related. This means that even if the
claimant’s pre-tort ‘chance’ of recovery had exceeded fifty percent he would not have had a
successful claim on traditional principles unless he could prove that he had suffered actionable
damage i.e. the harmful physical outcome. This means that Lord Nicholls was wrong to suggest
that ‘[t]he patient could recover damages if his initial prospects of recovery had been more than
50%’.43 The causation question is not the open-ended question of ‘what position would the
claimant be in if the defendant had acted carefully?’, but the focused questions of ‘has the
claimant suffered actionable damage and was the defendant’s negligence a cause of that damage?’
3.1.2 Causation
It is preferable to use the NESS test rather than the but-for test to address the causal link
because it enables the court to pinpoint more accurately the issues of proof that arise.
43
Gregg (HL) (n26) [2].
126
The advantages of the NESS test for causation are clear in Hotson because the NESS test treats
each cause as having equal significance since each was equally necessary for the sufficiency of the
set of conditions which resulted in the loss. As explained previously, avascular necrosis occurs
when insufficient blood vessels remain intact to keep the epiphysis alive. When the claimant first
went to hospital some vessels were damaged but some blood vessels remained intact although
distorted. The delay in commencing treatment meant that already ruptured blood vessels
continued to bleed into the joint thus increasing pressure in the joint and blocking the intact
blood vessels. The delay in diagnosing and treating the injury therefore caused damage to those
blood vessels that had remained and sealed the claimant’s fate by turning his injury into an
inevitability. When applying the NESS test the ‘sufficient set’ for avascular necrosis is
‘insufficient blood vessels intact to keep the epiphysis alive’. The relevant question is whether the
blood vessels damaged due to the delay were necessary for the sufficiency of this set. In other
words, when the claimant first went to hospital were there sufficient blood vessels remaining
intact for the epiphysis to stay alive?
Lord Bridge was quite clear that this was a problem of individual proof and did not involve any
deeper problem that would necessitate resort to statistics. He said ‘[i]n some cases, perhaps
particularly medical negligence cases, causation may be so shrouded in mystery that the court can
only measure statistical chances. But that was not so here. On the evidence there was a clear
conflict as to what had caused the avascular necrosis’.44 In other words, if the problem was more
akin to the evidentiary gap cases where there was a lack of understanding of the causal process
and the medical evidence could only furnish statistical assessments of risk then there may be a
reason for considering the loss of chance approach. But in this case the causal process was
understood, the problem was simply one of proof in the individual case and the claimant had not
managed to persuade the judge on the balance of probabilities.
Lord Nicholls sought distinguish the facts of Gregg from those of Hotson because there was:
44
Hotson (HL) (n4) 782.
127
…no significant uncertainty about what would have happened to Stephen Hotson’s leg if
treated promptly, once his condition at the time of the negligence has been determined
on the usual probability basis... Identifying Mr Gregg’s condition when he first visited Dr
Scott did not provide an answer to the crucial question of what would have happened if
there had been no negligence. There was considerable medical uncertainty about what
the outcome would have been had Mr Gregg received appropriate treatment nine
months earlier.45
The line that he draws between the degree and nature of the uncertainty in Hotson and in Gregg is
not a clear one. Yet regardless of whether proportional recovery would be an appropriate
exceptional solution, the ‘evidentiary gap’ in Gregg is not as deep as Lord Nicholls suggests. He
rightly observed that statistical evidence alone is unable to identify the cause in a particular case,
and statistical probabilities of over/under fifty percent do not translate directly into proof on the
balance of probabilities. But, as the next section will illustrate, statistical evidence can have an
important role in convincing a court that a causal link is more probable than not in a particular
case where the evidence itself is reliable and where it can be personalised sufficiently to the
claimant’s case. Medical understanding of the aetiology of cancer may be incomplete but a range
of risk factors are known so the statistical information may be personalised taking into account
the characteristics of the individual such as age, general health, spread of the cancer etc. Indeed,
Lord Phillips took a more rounded approach to the statistical evidence, examining its reliability
and making some, limited, attempt to personalise it. The focus of the next section is on the role
that statistical evidence can play in informing the balance of probabilities standard of proof
because the courts need to understand this fully before they can identify accurately the problems
that are raised by claims such as Hotson and Gregg.
45
Gregg (HL) (n26) [38].
128
3.2
The balance of probabilities standard of proof
It is important to remember that the balance of probabilities standard of proof does not refer to
‘probability’ in the statistical sense, but refers to the degree of belief that must be established in
relation to the fact of causation. It is a qualitative assessment that may be informed by statistical
or epidemiological evidence but which ultimately must also incorporate an assessment of the
reliability of the evidence and of how far it can be personalised to the individual claimant.46
The balance of probabilities is often treated as a statistical tool, where causation is proven as
soon as the 51% threshold is passed. In Gregg, Lord Nicholls began his speech by stating that
‘[t]he patient could recover damages if his initial prospects of recovery had been more than 50%.
But because they were less than 50% he can recover nothing’.47 This is also common in academic
commentary, for example Beever has suggested that ‘[t]he problem facing the plaintiff was that,
before the defendant’s negligence, his chance of being cured was only 42%. This meant that, on
the balance of probabilities, the defendant did not deprive the plaintiff of a cure’.48 Yet as Lord
Nicholls observed:
Statistical evidence, however, is not strictly a guide to what would have happened in one
particular case. Statistics record retrospectively what happened to other patients in more
or less comparable situations. They reveal trends of outcome. They are general in nature.
The different way other patients responded in a similar position says nothing about how
the claimant would have responded.49
He continued to ask,
Who can know whether Mr Gregg was in the 58% non-survivor category or the 42%
survivor category? There was no evidence, peculiar to him or his circumstances, enabling
Epidemiology is the branch of medicine that studies the incidence and causes of disease in populations.
Gregg (HL) (n26) [2].
48 Beever, ‘Gregg v Scott’ (n39) 201. See also Stapleton ‘Loss of the chance of cure’ (n10) 997; Green, ‘Coherence of
medical negligence cases’ (n39) 3.
49 Gregg (HL) (n26) [28].
46
47
129
anyone to say whether on balance of probability he was in the former group of the latter
group.50
It will be argued later that there was evidence enabling the court to say which group the claimant
belonged to, but it is important to note that Lord Nicholls is right to say that raw statistics do
not tell us anything about the individual. This means that if the raw statistics were reversed and
58% of patients would have survived, the court would still be unable to say whether the claimant
fell into the survivor or non-survivor category if there was no information enabling them to
personalise the statistics.
This tendency to map the patient’s statistical chance directly onto the balance of probabilities is
perhaps more common in cases of medical negligence because a doctor explaining a patient’s
prognosis will routinely refer to the statistical likelihood of recovery. A purely statistical approach
may mean that where the patient’s chance of recovering is very close to 50 percent the case will
be seen as being very close to the margins. But once he has suffered the harmful outcome then it
is a question of fact whether the condition was treatable, and the balance of probabilities is a
matter of belief in this fact and is designed to deal with the uncertainty – the court does not need
to be convinced that the injury was certainly treatable, it needs only be convinced that it is more
likely than not that it was treatable. It is essential to understand the role that statistical evidence
can play in informing the balance of probabilities standard of proof.
The balance of probabilities is a standard of proof or persuasion, so it is a ‘belief probability’.
Barnes explains that belief probability refers to ‘the credibility – the believability – of the
evidence in support of a party’s factual claims’.51 He distinguishes this from ‘fact probability’ and
‘sampling error probability’.52 Fact probability refers to the likelihood of a causal relationship.
One aspect of this is the ‘risk ratio’ or ‘relative risk’:
Gregg (HL) (n26) [29].
David W Barnes, ‘Too Many Probabilities: Statistical Evidence of Tort Causation’ (2001) 64 Law and Contemp.
Probs 191, 192.
52 See also Steve Gold, ‘Causation in Toxic Torts: Burdens of Proof, Standards of Persuasion, and Statistical
Evidence’ (1986) 96 Yale LJ 376.
50
51
130
Risk ratios measure the percentage change in the incidence of a specified harm, such as a
disease. A risk ratio compares a background rate, where the stimulus in question is not
present, to the rate that obtains when the stimulus is present. For example, in a routine
tort case alleging that a negligent failure to light a stairway caused a fall, a risk ratio might
compare the incidence of falling down stairs when the stairs are well-lit to the incidence
of falling when the stairs are unlit…A risk ratio greater than one indicates that risks are
increased. For instance, risk ratios of 1.5 and 3 indicate that the stimulus (for example,
lack of lighting) increases the risk of falling by 50% and 200% respectively.53
The ‘sampling error probability’ refers to ‘a statistical property of data underlying evidence
offered to prove a relevant fact’,54 so it aids the assessment of the reliability of a particular fact
probability. Barnes explains that:
Even when a sample is composed of randomly chosen subjects, those subjects may not
represent accurately the population. That possibility means that any observed statistical
relationship between acts like the defendant’s and harms like the plaintiff’s revealed by a
study of a sample may be due to the happenstance of having drawn randomly an atypical
sample.55
Barnes explains that in all studies based on a sample of the population there will be a sampling
error and that this ‘is not an error in the design of the sample. Indeed, it is not an error
attributable to any person. It is an unavoidable property of inferential statistics, the process of
estimating attributes of a population by examining a sample’.56 The sampling error probability is
expressed as a ‘p-value’ ranging from 0 to 1.00, and a p-value closer to 0 indicates a smaller
probability that error is due to the make-up of the sample. The greatest sampling error
probability that is accepted in science is five percent, meaning that on the basis of sampling error
Barnes (n51) 193.
ibid.
55 ibid.
56 ibid 198.
53
54
131
there must be at least a 95 percent chance that the relationship is causal and not due to the
chance that the sample is unrepresentative of the population. Furthermore:
A sampling error probability may as easily be calculated from a poorly designed study as
from a randomized, controlled, double-masked study. The credibility of that probability
and any fact probability derived from that study, however, depends on the quality of the
design study.57
In other words, ‘[t]he p-value does not measure the probability that the design was perfect;
rather, it assumes the design was perfect’.58
This means that in order for the court to form a rational belief in causation based on statistical or
epidemiological evidence the reliability of the study must also be evaluated by addressing factors
such as the experimental design and measurement. There are a range of criteria available for
assessing the reliability of an epidemiological study and the statistics to which it gives rise. Barnes
explains that the ‘well-known gold standard for experimental design is the randomized,
controlled, double-masked study’.59
After the quality of a study has been established, epidemiologists then need to test the reliability
of the causal inferences that can be drawn from the data. Miller has explained, ‘[i]n the UK, the
nine criteria articulated by Professor Bradford Hill are perhaps the best known’60 criteria to test
causal inferences. These are: the strength of association, how high is the relative risk?;
consistency, have the results been replicated elsewhere?; specificity of cause and of effect, does
the ‘cause’ produce only one effect? Does the ‘effect’ have only one cause?; temporality, the
effect must follow the cause; biological gradient, is a ‘dose relationship’ identifiable?; plausibility,
is it consistent with prevailing biological knowledge?; coherence, does it conflict with any known
biology of the disease?; experiment, does the frequency of the effect fall when the ‘cause’ is
ibid 200.
ibid 204.
59 ibid 200.
60 Chris Miller, ‘Causation in personal injury: legal or epidemiological common sense?’ (2006) 26 LS 544, 545.
57
58
132
removed?; analogy, is the cause-effect relationship of any comparable disease accepted?61 These
factors relate to the question of whether there is a general causal relationship between the ‘cause’
and ‘effect’ i.e. the harmful agent for which the defendant is responsible and the type of illness
suffered by the claimant. Miller notes that ‘Hill took pains not to exaggerate the claims of his
criteria: “None of my nine viewpoints can bring indisputable evidence for or against the causeand-effect hypothesis and none can be required as a sine qua non”’.62 While none of the factors
alone is determinative of the quality of the study, together they enable its reliability to be
evaluated.
Furthermore, in order to form a rational belief about the causal relationship in an individual case,
the court must assess the extent to which it is possible to extrapolate from the fact probability
about general causation to causation in the individual case. This involves other factors, notably
how closely matched the claimant is with the sample population and with the general population.
So even if a study is reliable, the sampling error probability is not the same as a belief probability
because it ‘would be divorced from the belief probability just where we need it – where we ask
whether the particular defendant’s act was a necessary event in producing the particular plaintiff’s
harm’.63 All of these factors therefore affect our overall degree of belief in the fact that the
particular defendant’s negligence was a cause of the particular claimant’s loss, the belief
probability. This means that statistical evidence can provide a starting point for forming a
rational belief in causation in an individual case, but it must be interrogated for reliability and for
how far it is capable of informing us about the individual claimant.
It is important not only for lawyers to understand what the ‘balance of probabilities’ entails but
also to ensure consistency in the discourse between lawyers and medical experts. The trial judge
in Hotson noted that the claimant’s expert witness ‘speaks of likelihood as something involving a
ibid 548.
ibid 548. Citing A Bradford Hill, ‘The environment and disease: association or causation?’ (1965) 58 Proc R Soc
Med 295.
63 ibid 204.
61
62
133
less than 50% chance in contradistinction to a probability as denoting more than that’.64 In other
words, the medical expert drew a significant distinction between something being ‘likely’ and it
being ‘probable’. Given the importance of the ‘balance of probabilities’ in determining the
outcome of a case it is essential that all parties understand the concept and use the term
‘probability’ consistently.
Indeed the pressure from the lawyers to express the chances in statistical terms seemed to go
against the medical experts’ position in Gregg which was that a figure could not be put on the
individual chance. With regard to the claimant’s pre-tort chance of cure, Inglis J noted:
Professor Goldstone was unable to put a percentage figure on that chance. It has to be
taken as less than evens...It might be said that since neither he nor Dr Bunch put a figure
on it, the court should not. I do not agree...The experts thought it possible that his
individual prognosis had been reduced to less than 50% of what it would have been
intrinsically at the outset.65
So since the balance of probabilities involves a qualitative assessment of the evidence it is
important not only to avoid the temptation to map statistical probabilities directly onto it but
also to avoid pressuring expert witnesses to quantify the risk if they are unable to.
3.3
Proof in Hotson
The details of the decision at first instance in Hotson show clearly that the judge did undertake a
detailed qualitative assessment of the conflicting evidence. 66 That evidence did not consist simply
of conflicting statistical probabilities of avascular necrosis following a fall. The 25% ‘chance’ was
not a quantitative assessment of the likelihood of harm, but was a numerical expression of the
judge’s degree of belief in the fact that the injury could be avoided; it reflected his qualitative
assessment of the evidence. He looked at how severe the displacement of the bone was, how
Hotson v East Berkshire HA [1985] 3 All ER 167 (QBD), 173.
Para [51] (cited in Gregg (HL) (n26), [164] (Lord Phillips)).
66 This is set out in the All England Law Reports (n64) but unfortunately is omitted from the Weekly Law Reports.
64
65
134
easily the facture was reduced, the different interpretations of the x-rays, disagreement over
whether the blood vessels would have continued to bleed and build up pressure on those
remaining intact or would have sealed themselves, a paper from a medical journal that
considered different treatment options, and a paper that analysed nine cases of fracture
separation of the epiphysis in children. He found that there were internal inconsistencies in the
evidence of the claimant’s expert witness as well as between the two conflicting experts. On one
issue he said ‘I am not persuaded by the published material before me…Rather I am on balance
persuaded to the contrary view’.67 Eventually concluding:
I regret that I found certain parts of the evidence of both experts, highly qualified and
experienced although they both undoubtedly are, difficult to accept, either as a result of
internal inconsistency within their evidence or because of what seemed to be an intrinsic
want of logic in some particular expressed view…In the result I find myself unattracted
to, and finally unable to accept, either of the competing extreme views.68
This was the basis for his finding that it was ‘possible but improbable’ that the fall had ‘left intact
sufficient vessels to keep the epiphysis alive’.69 Whilst he expressed the ‘chance’ as a percentage,
it is clear from the language used throughout the judgment that this 25% ‘chance’ was not a
statement of how likely it was, given the number of blood vessels remaining intact, that the
claimant could be cured, but a statement of how likely it was that enough vessels remained intact
for him to be cured. It therefore reflected his degree of belief in the proposition that enough
blood vessels remained intact for the epiphysis to be kept alive. It was expressed numerically to
allow the calculation of damages at the valuation stage. Normally, if approaching this as a
conventional claim for the physical harm, the judge would not need to put a figure on the extent
of his doubt. The balance of probabilities merely requires him to find, as he did, that on balance
he was persuaded that insufficient vessels remained intact. But because he was willing to
Hotson (QBD) (n64) 173 (emphasis added).
ibid.
69 ibid 171.
67
68
135
approach it as a loss of a chance case, he has quantified the extent of the remaining doubt in his
mind as being 25% so that he could then calculate a proportionate award of damages - this is in
no way the same as saying that the doctor created a 25% risk, or caused the claimant to loss a
25% chance, or contributed 25% of the chance of harm.
3.4
Proof in Gregg
In Gregg the expert evidence at first instance consisted of a joint report by a consultant
haematologist Professor Goldstone (instructed by Mr Gregg) and Dr Bunch (instructed by Dr
Scott) concerning the claimant’s disease and the effect of the delay in diagnosis, and oral
evidence by Professor Goldstone alone. In the period between the claimant’s pleading of the
particulars of the claim in April 2000 based on Professor Goldstone’s evidence, and the writing
of the joint report in 2001, Professor Goldstone became aware of a paper that had been
published in 1999 by the American Society of Hematology, written by Falini et al (the ‘Falini
paper’) which suggested that the particular type of lymphoma could be further divided into two
sub-categories, ALK positive and ALK negative. Those who fell into the ALK negative subset
had a significantly worse prognosis than those in the ALK positive category. The claimant’s
lymphoma was ALK negative. The joint report by Professor Goldstone and Dr Bunch took
account of the Falini paper. Only Professor Goldstone gave oral evidence and developed a
‘working example’ of the fate of a cohort of 100 ALK negative patients based on the evidence in
the joint report. Latham LJ noted that ‘[t]he assumption was that the cohort consisted of patients
with the same stage of disease as that from which the appellant suffered at the time when
treatment should have been commenced’.70
The working example showed:
Of those with initial treatment by CHOP chemotherapy with or without field
radiotherapy 55 will achieve complete remission. 45 will not achieve complete remission,
70
Gregg (CA) (n28) [14].
136
and of those 41 will then die. Four who did not achieve complete remission immediately
will be brought to achieve it by further treatment of various kinds. Thus of the initial 100
59 manage to achieve complete remission. Of those 35 do not relapse. They are
described by Professor Goldstone as the core group of survivors. 24, however, do
relapse (usually, if they are going to, within two years or so of achieving remission). Of
those 24 half, a further 12, will not be responsive to further treatment and will die. The
remaining 12 will be responsive to further treatment, typically high dose chemotherapy
with stem cell harvesting such as the claimant himself went through. Of those 12 half, a
further six, will not relapse again and will become survivors. Of the remaining six who do
relapse again, only one will survive. The number of survivors from the 100 who started
out will therefore be 42.71
During the period of delay following the negligence the cancer upstaged from stage 1 (presence
in the lymph nodes) to stage 1E (spread to other tissues outside the lymph nodes). There appears
not to be a similar working example for a cohort of 100 ALK negative patients at stage 1E,72 but
the expert evidence in the joint report was that ‘it was “quite possible” that his individual
prognosis had been reduced to less than 50% of what it would have been intrinsically at the
outset’.73 At the time when the disease was actually diagnosed and treated the likelihood of cure
was ‘no more than 10-15%’,74 but at the time of trial ‘[h]is actual prognosis they considered to be
dependent now more on the fact that he had survived for three years’ which meant that his
chances of ten years survival ‘had increased from a statistical 10% to 20-30%’.75 Thus ‘the
Statement of Facts and Issues agreed by the parties included a statement that the judge found
that the effects of the delay in commencing Mr Gregg’s treatment meant that his “chances of
Findings at first instance, cited in Gregg (HL) (n26) [146] (Lord Phillips).
As will be discussed later, this is because the working model was actually based on patients at all stages of the
disease (see text to (n82)).
73 Gregg (CA) (n28) [13] (Latham LJ).
74 ibid.
75 ibid [15].
71
72
137
long-term survival fell as a result of the negligence from 42% (on initial presentation to [Dr
Scott]) to 25% (as at the date of trial)”’.76
As noted at the start of this section, Mr Gregg had not suffered the harmful outcome to which
this ‘chance’ related since he was still alive. Had he died, the causation question would be
whether the negligent delay in diagnosis was a cause of his death, so it should be established on
the balance of probabilities that his disease could have been cured. There are two areas of
concern in this case: first, the reliability of the evidence; second, the court’s failure to personalise
the statistical evidence to the claimant.
3.4.1
Reliability of the evidence
Lord Phillips’ examination of Professor Goldstone’s evidence led him to conclude that it was ‘a
very inadequate tool for assessing the effect of the delay in treatment on Mr Gregg’s progress
and prognosis’.77 McIvor has identified a number of concerns regarding Professor Goldstone’s
evidence which affect the degree of belief that can be placed in the statistics he presented to the
court.78 She notes that as a consultant haematologist his expertise is medical rather than statistical
or epidemiological so whilst skilled in diagnosis and treatment of certain illnesses, he is less
skilled in the interpretation of data. She further highlights that he had initially suggested that the
statistical probability of survival was 84% because he had not taken into account the paper by
Falini et al which showed that ALK negative patients had a much lower chance of survival than
ALK positive patients. This paper had been available at the time of writing of his first report so
its omission casts doubt on the quality of his evidence.79 The Falini paper itself was based on a
very small sample consisting of only 88 patients, 53 of whom were ALK positive and only 25 of
Gregg (HL) (n26) [147] (Lord Phillips).
Gregg (HL) (n26) [157].
78 Claire McIvor, ‘The Use of Epidemiological Evidence in UK Tort Law’ in S Loue (ed.) Global Perspectives in Forensic
Epidemiology (forthcoming).
79 ibid.
76
77
138
whom were ALK negative.80 Given the small sample size the accuracy of this data is probably
limited. Furthermore, the ‘working example’ that Professor Goldstone devised to predict the
likely fate of 100 ALK negative patients seems to be based on his own experience. Presumably
given that he had only recently become aware of the Falini paper, and therefore was only recently
averted to the reduced survival rate of ALK negative patients, it can only have been a rough
model. This does not suggest that the statistics he produced are especially reliable, it suggests that
they are the best that he could do based on his clinical experience and that a different expert
could have given different statistics.
3.4.2
Personalising the evidence
Even if this evidence was reliable, in order to form a rational belief as to whether the claimant’s
disease could have been successfully treated it is essential to personalise it to the individual as far
as possible. Professor Goldstone had made some attempt to personalise the statistics to Mr
Gregg’s post-tort chances in light of the progress of his treatment, suggesting that the likelihood
of survival had increased from 10-15% at the time of diagnosis to 20-30% at the time of trial.
But there was no corresponding attempt to personalise the statistics in relation to his pre-tort
chance of survival despite there being information on which this attempt could have been based.
Stapleton explains this argument:
The unpersonalized estimate of [Mr Gregg’s pre-tort chance of being cured] was 42%.
But by the time of trial we had an important piece of information about the actual
experience of this particular claimant after the commencement of treatment, which could
have been used to “personalize” that estimate. By the trial we knew that, even after
breach, Mr Gregg was able to achieve complete remission in 1996. Does this not show
that, a fortiori, had there been no breach, Mr Gregg would have been one of the 59 out of
80
Gregg (CA) (n28) [11].
139
100 who initially achieved complete remission? A member of that group had a pre-tort
chance of cure of 42/59 = 71%, not 42%.81
It was noted above that it was assumed that Professor Goldstone’s working example applied to a
cohort of 100 patients whose disease was at the same stage as the claimant’s pre-tort disease, and
that he did not provide a similar working example for 100 patients whose disease was at the
more advanced stage that Mr Gregg’s cancer had reached when treatment commenced. It is thus
unclear where the assessment of the claimant’s post-tort likelihood of cure at 10%-15% was
derived from. Furthermore, Lord Phillips questioned this assumption, explaining ‘[w]hen
describing Professor Goldstone’s model, the judge stated: “The 100 patients in the worked
example include all ages, and also people with other unrevealed personal characteristics, one of
which is the stage of the disease at diagnosis”’.82 This means that the court could have made some
attempt to personalise the statistics by considering the claimant’s age, and the stage of his
disease. Surely, given the early stage of his disease he was more likely to belong to the survivor
category than somebody whose disease is more advanced?
So in any negligence action the court must seek to personalise the evidence to the individual
claimant in order to form a rational belief in the likely causes of the damage, and should also
evaluate the reliability of the evidence. Mr Gregg’s claim would have been hindered because he
had not suffered actionable damage in the form of the final outcome, and because the evidence
pertaining to causation had not been personalised and in any case was not reliable. It seems
unlikely that he would have been able to prove causation on the balance of probabilities even if
he had suffered actionable damage, but there is nothing to suggest that this is a particularly
unusual problem requiring an exceptional solution. The claimant’s argument was that the
application of conventional principles resulted in an unfair failure of his claim, so exceptionally
fairness demanded that the loss of a chance of a better outcome should be substituted as the gist
of the action. This section has demonstrated that there were no inherent obstacles to proof of
81
82
Stapleton, ‘Loss of the chance of cure’ (n10) 1000.
Gregg (HL) (n26) [148].
140
causation that would have made rejection of his claim ‘unfair’, it is simply that the evidence in his
case was weak, so there seems to be no reason why ‘fairness’ should have a claim on the law in
preference to corrective justice. The next section turns to the notion of ‘chance’ itself to consider
whether it could form the gist of a negligence action if fairness so required.
4.
‘Loss of a chance’ as damage
In this section it will be argued that in the context of medical negligence, specifically the
misdiagnosis or mistreatment of an existing illness, the loss of a chance ought to be recognised
as damage that can form the gist of a negligence action. Instead of approaching the question of
loss of a chance as a way of assisting a claimant who faces difficulties of proof of causation,
however, the issue will be approached from the perspective of whether a lost chance can
constitute actionable damage. This means that ‘chance’ will be examined in its own right, rather
than as leading to recovery of damages that are proportional to the physical harm to which the
chance relates. In other words, it will be argued that a lost chance is a distinct form of damage
that ought to have a place in the tariff set out in the Judicial College’s Guidelines for the Assessment
of General Damages in Personal Injury Cases.83 This will therefore continue to develop the argument
that the issues of damage, causation and quantification are three distinct issues, and that clarity in
the law can only be achieved by having a full understanding of the content and role of each
concept and the interrelationship of the three.
This section will begin by addressing the concept of chance in order to demonstrate that, in the
limited context of medical misdiagnosis/mistreatment of existing illness, a ‘chance’ is something
of value that can be lost and ought to be recognised as being an actionable form of damage.
There is a unique relationship between doctor and patient when treating an existing illness since
the outcome of treatment is uncertain for both, meaning that both parties value the patient’s
chance of being cured. As argued in Chapter One, these issues of interpersonal responsibility
Judicial College, Guidelines for the Assessment of General Damages in Personal Injury Cases (11th ed., Oxford University
Press 2012)
83
141
present in the doctor’s duty of care should be allowed to shape the negligence inquiry so that loss
of a chance can be recognised as damage. This is consistent with a corrective justice based
approach to negligence which requires the defendant to repair wrongful loss that he caused. An
important aspect of recognising loss of chance as actionable damage is the argument that the
concept of the loss of a chance is distinct from the concept of an increase in risk, so recognition
of claims for loss of a chance in this limited context would not have the expansionary effect that
is often feared.84 Finally the issue of valuing this chance will be addressed in order to show that
since the chance is a distinct loss its value is independent of the final harm that the patient
wishes to avoid. It will be clear that loss of a chance is not being presented as a solution to a
problem of causation, and that this in turn will lend clarity to future causal problems in medical
negligence.
4.1
Rejecting ‘chance’ conceived as a proportion of the physical outcome
The House of Lords in Hotson avoided the question of whether ‘loss of a chance’ was actionable
because it was held that the patient did not have a chance of recovery when he first went to
hospital. Instead, it was held, he either had sufficient blood vessels remaining undamaged and
treatment would have succeeded, or insufficient blood vessels remained intact and he was
doomed to develop avascular necrosis. Likewise, in Gregg it was held that the forty-two percent
chance of recovery meant that in a cohort of 100 patients, 42 would recover and 58 would not.85
The claimant either was one of the 42 or one of the 58, but he did not personally have a ‘chance’.
Such arguments are broadly based on the idea that the physical process of disease is
‘deterministic’ in nature. It is this issue of determinism of causal processes that will be examined
in further detail. These cases were correctly decided. The fact that the claimants were unable to
prove causation but put forward the loss of chance argument seeking to recover a proportion of
the value of the relevant physical harm shows that they were effectively asking to be
84
85
See for example Michael Jones, ‘Proving Causation – Beyond the ‘But For’ Test’ (2006) 22 PN 251
Assuming these figures were based on reliable evidence.
142
compensated for the mere possibility that the defendant had caused them physical harm. As the
previous section showed, the ‘chance’ was actually a quantification of the judge’s doubt over the
causal link. This is inconsistent with corrective justice where a wrongdoer is required to repair a
loss because he caused it through his wrongdoing.
Following the decision in Hotson, the concept of chance was elaborated in an article by Helen
Reece, 86 in which she sought to reconcile that decision with the decision in Chaplin v Hicks.87 In
Chaplin, the claimant was a woman who had entered a beauty contest organised by the defendant.
Entrants were narrowed down to a final group of 50 women who were then invited to an
appointment with the defendant who would select 12 winners from this group. The claimant was
unable to keep her appointment and sued the defendant for breach of contract for failing to give
her a reasonable opportunity to present herself for an appointment. The defendant argued that
damages could not be quantified, but the Court of Appeal held that since the claimant had
roughly a one in four chance of winning (i.e. 12 out of 50 chance) she should be awarded 25
percent of the value of the prize that she would have received had she won the competition.
Reece argues that the distinction lies in the causal process itself which was ‘deterministic’ in
Hotson but ‘(quasi) indeterministic’ in Chaplin. By ‘deterministic’ she means the ‘hypothetical
chain of events is fully determined by the events which have occurred’.88
This means that in a deterministic world, given sufficient knowledge the cause of anything could
be discovered and the future could be predicted with certainty. In contrast, if a process is
indeterministic it cannot be predicted even with unlimited knowledge. Reece explains that in an
indeterministic process, probability is an objective concept. The likelihood of a future event
occurring has an objective probability that is a property of the natural world. In contrast, in a
Reece (n23).
Chaplin v Hicks [1911] 2 KB 786 (CA).
88 Reece (n23) 192. She provides the following explanation from Laplace: ‘We ought then to regard the present state
of the universe as the effect of its anterior state and as the cause of the one which is to follow. Given for one instant
an intelligence which could comprehend all the forces by which nature is animated and the respective situation of
the beings who compose it – an intelligence sufficiently vast to submit these data to analysis – it would embrace in
the same formula the movements of the greatest bodies of the universe and those of the lightest atom; for it,
nothing would be uncertain and the future, as the past, would be present to its eyes’ (Laplace, Philosophical Essay on
Probabilities (1819, Springer-Verlag English translation 1995) 4).
86
87
143
deterministic process there is no objective probability of something occurring – either it will or it
will not and theoretically this can be predicted with certainty. So when probability is used to
describe the chance of such an event, the probability is not objective but epistemological; it is an
expression of the likelihood of an event given the limited knowledge that is available. Reece
explains, for example, that in an accident where a motorist drove into a pedestrian who then
suffered broken bones, the bones broke because the force of the car was applied to them ‘and
bones do not break in the same way on a random basis’.89 Similarly, given sufficient knowledge it
would be possible to predict with certainty whether or not a pedestrian’s bones would be broken
in any particular case. So when road safety advertisements say that if a driver hits a child at
40mph there is an 80 percent chance that the child will die, but at 30mph there is an 80 percent
chance that the child will live, this is an epistemological assessment of the ‘chance’ that the child
will die, but in any particular case the outcome (life or death) would be determined by the precise
conditions. Given sufficient knowledge, in the instant that a child is hit by the car it would be
possible to say with certainty that this child will live or die. In contrast, the decay of a Uranium
atom is random, and similarly the outcome of a lottery is random and cannot be predicted so the
holder of a lottery ticket has an objective chance of winning.90 In these indeterministic cases,
Reece concludes, the balance of probabilities standard of proof simply cannot cope so the courts
turn to a loss of chance approach which reflects the underlying existence of an objective
probability.91 So she rejects the notion of ‘loss of a chance’ as damage unless there was an
objective chance.
Reece argues that the causal process involved in Hotson was deterministic:
[T]here was a time in the past when the cause of the necrosis could have been
determined. If the blood vessels had been examined after the fall, then it would have
been humanly possible to decide whether or not the plaintiff would develop necrosis
Reece (n23) 195.
ibid.
91 Reece (n23) 204-5.
89
90
144
even if he were treated. At the time of the trial the cause was uncertain, but the
uncertainty was epistemological not objective.92
Since the claimant did not have an objective chance of recovery, Reece concludes that the ‘loss
of a chance’ cannot form the gist of the negligence action. It is clear that once a claimant has
suffered physical harm the causes of this are fixed because the physical process was
deterministic, so there is not a ‘chance’ that the defendant was a cause there is merely uncertainty
as to whether he was a cause. This means that the loss of chance argument, as it was presented in
Hotson and in Gregg, cannot be accepted.
Understanding that the ‘chance’ in these cases was an epistemological probability explains why
the patient’s ‘chance’ fluctuated in Gregg. The House of Lords was reluctant to accept the loss of
chance argument in Gregg because of the apparent instability of the ‘chance’ since the statistical
probability of the claimant being cured fluctuated over time as his treatment progressed. It is
clear that this is more than simply a practical concern about the date at which the ‘chance’ should
be measured if the claim had been allowed, but reflects the underlying fact that he did not have
an objective chance. The ‘chance’ is an epistemological probability, meaning that it reflects the
uncertainty over the outcome, so as the patient gets closer to achieving an outcome the
uncertainty decreases so his ‘chance’ will get closer to 0% or 100%.
4.2
Epistemological chance as an independent form of damage
As argued above, once the claimant has suffered physical harm the concept of loss of a chance is
generally rejected on the basis that a patient does not have a ‘chance’ as such. Either he was
destined to recover, or he was destined not to, but he did not have a ‘chance’ of recovering.
Rather, the ‘chance’ is an epistemological probability, that is it is an expression of our belief in
the likelihood that the patient is one of the people destined to recover based on the limited
knowledge available. The argument that will be developed here is that in the face of uncertainty
92
ibid 196.
145
at the time of diagnosis and treatment both the patient and the doctor treat the epistemological
chance as though it were an objective chance. There is a unique relationship between doctor and
patient when treating an existing illness since the outcome of treatment is uncertain for both,
meaning that both parties value the patient’s chance of being cured. These issues of interpersonal
responsibility present in the doctor’s duty of care mean that the law should adapt to the reality of
that uncertainty rather than striving towards a philosophically correct approach.
It is important to remember at the outset of this section that what is under consideration is
simply the question of whether the patient’s chance of recovery is something that is capable of
constituting loss. This should be kept separate in the mind from the question of whether the
patient also suffers physical harm because this is not an attempt to assist a claimant who is facing
difficulties of proof of causation. It should also be kept separate from the question of valuation
of the chance when assessing damages.
4.2.1
Post-diagnosis factors that determine the outcome
Even if the physical process is deterministic the outcome of treatment will often be beyond the
control of the doctor who cannot guarantee a cure, and this is particularly true when the
outcome is affected by factors beyond his control that arise after the diagnosis. In other words,
although the causal process may be deterministic, the outcome is not always determined at the
time of diagnosis. Things such as the patient’s diet and lifestyle may affect the course that the
treatment takes, so the doctor regards the treatment as having a chance of success.
The extended time period for treatment of the disease in cases such as Gregg means that medical
advances may be made over the course of the treatment that improve the prospects of cure as
Lord Nicholls noted when he said ‘[n]ew and improved drugs and procedures make possible
ever more alleviation of illnesses and injuries’.93 In the Court of Appeal, Mance LJ explained:
93
Gregg (HL) (n26) [23].
146
[T]he most obvious influencing factors are, one would suppose, internal to the claimant
at the time of the negligence, however unknown or unknowable they may be; and they
consist of the precise characteristics, development and spread of the cancerous cells at
the time of the negligence as well as the claimant’s precise physical characteristics and
resistance. Other influencing factors may very well include subsequent events such as the
particular medical treatment received, the patient’s subsequent lifestyle and his or her, or
indeed others’, reaction to the stress inevitably incurred.94
This illustrates that even if the causal process of cancer is deterministic, the outcome was not
necessarily determined at the time of the negligence. If subsequent factors such as the patient’s
lifestyle and his response to stress are factors that determine the outcome then it is not possible
to say whether, at the time of the negligence, Mr Gregg was fated to die or destined to recover.
The possible impact of stress upon the course of his disease is especially relevant because the
delay in diagnosis and consequent knowledge that his cancer had spread, which meant that he
was statistically more likely to die, will have affected the patient’s physical and mental state. This
means that the negligence will have contributed to a factor (stress) which is one of the factors
that determines whether or not he will be cured.
This suggests that a patient’s physical state at the time of negligence was not, on its own,
determinative of the outcome so a doctor could not even theoretically predict with certainty
whether the treatment would lead to a cure. This inherent uncertainty means that he considers
the patient to have a chance, and the law should reflect this reality in recognising a loss of that
chance as damage. However this chapter seeks to make a more far-reaching argument that even
where the outcome is determined at the time of the doctor’s negligence, as in Hotson, at the time
of the negligence both the patient and the doctor regard the prognosis as uncertain and therefore
consider the patient to have a chance of recovery. This means that any patient whose illness is
misdiagnosed or mistreated ought to be considered to have lost a chance if that delay affects the
94
Gregg (CA) (n28) (Mance LJ). See also Gregg (HL) (n26) [80] (Lord Hoffmann).
147
medical assessment of the likelihood of cure. In this model the final physical harm is still seen as
having a determinate cause which ought to be established on the balance of probabilities. The
loss of chance argument for proportionate recovery has no place in this backwards-looking issue.
However, if we situate ourselves at the time of the negligence, looking forward, the outcome is
uncertain in the eyes of both the doctor and the patient, so both parties regard the patient as
having a chance of cure at that time, and any damage to this chance is a distinct form of harm
deserving of compensation. Since this harm is distinct from the final outcome its value is also
independent of the final outcome rather than being valued as a proportion of the physical harm.
Because it is an independent form of harm this also means that it would be actionable even if
ultimately the harm materialises.
4.2.2
The grey-areas on the deterministic-indeterministic spectrum
In her article on loss of a chance, Reece adopted a working understanding of determinism and
indeterminism which showed two extremes which enabled her to illustrate the different types of
causal process, but she reached the more nuanced conclusion that ‘a continuum between
determinism and indeterminism is more plausible than a clear division’.95 It will be argued in this
section that medical negligence cases involving misdiagnosis or mistreatment of existing illness
do indeed fall somewhere in the middle of this spectrum. The physical causal process may well
be deterministic, so once physical harm has occurred it is theoretically possible to discover the
causes of this outcome. At the time of diagnosis it is also theoretically possible, with unlimited
knowledge and understanding, to predict the outcome with certainty. But crucially at the time of
diagnosis and treatment the practical reality is that both the claimant and the defendant doctor
have only limited knowledge and understanding so they can only state the likelihood of a cure.
Although they may be aware that the philosophical/scientific truth is that this is an
epistemological probability, they both treat the patient as having a personal chance of cure. The
95
Reece (n23) 206.
148
patient may adapt his lifestyle to minimise risk factors and improve his chance of cure, the focus
of the doctor’s duty of care is on treating the risk factors (i.e. the illness) in order to maximise the
patient’s chance of being cured. This characteristic of the doctor’s duty of care means that the
demands of interpersonal responsibility favour the recognition of this form of damage in order
to achieve a coherent approach. The causal process is deterministic but in the context of this
special relationship both parties treat it as being indeterministic, indeed the purpose of the
defendant’s duty of care is premised on it being indeterministic. This means that the damage to
this epistemological ‘chance’ is a loss that ought to be repaired under corrective justice.
Reece acknowledges that it is difficult to conclude that an event is indeterministic because ‘our
inability to predict its occurrence could be either because the event is inherently unpredictable or
because we have not yet found the complete set of necessary and sufficient causes’.96 She is
willing to characterise an event as ‘quasi-indeterministic’ i.e. to the best of our knowledge
indeterministic:
if and only if it could not have been predicted at any point in the past, it cannot be
predicted in the present even given unlimited time, resources and evidence, and we
cannot imagine how it would become predictable in the future, even given the success of
present research programmes.97
This is clearly a very demanding definition. It sets a high threshold and means that most physical
processes will be regarded as deterministic.
It is clear that a grey area exists if one compares the views of different authors regarding an event
as familiar as the toss of a coin. Reece treats this as indeterministic, stating that ‘[t]ossing a coin
does not cause the coin to land heads, but it does cause the chance that the coin will land
heads’.98 In contrast, Beever regards it as deterministic, explaining ‘[w]hen the coin is in the air
there is a fact of the matter as to which side it will land on. There is, then, no objective chance
ibid 194.
ibid 194.
98 ibid 206.
96
97
149
here whatsoever’.99 This highlights the fact that where it is theoretically possible to predict an
outcome, but impossible to predict it given normal knowledge or experience, it is common to
treat an event as being indeterministic.
Reece’s definition of determinism may be an acceptable theoretical and philosophical definition
but it does not reflect the experience of the doctor and the patient. Even with his medical
knowledge and understanding, the doctor is unable to predict the outcome with certainty, instead
describing the patient as having a chance of recovery and then treating the illness. Providing
appropriate treatment is the way the doctor finds out whether the patient could or could not be
cured. As Lord Nicholls argued, the doctor regards the patient as having a chance of being
cured, explaining:
the law should be exceedingly slow to disregard medical reality in the context of a legal
duty whose very aim is to protect medical reality. In these cases the doctor’s duty to act
in the best interests of his patient involves maximising the patient’s recovery prospects,
and doing so whether the patient’s prospects are good or not so good.100
In a case such as Hotson the process is deterministic, and medical science has advanced enough to
understand that it is the state of the blood vessels that determines whether the damage will
occur. But, in any individual case, when the doctor commences treatment both he and the
patient regard the outcome as involving a chance because they simply do not know whether
sufficient blood vessels remain intact. Even with careful treatment the doctor cannot guarantee
that the treatment will succeed or fail because there are factors beyond his control and because
he does not know whether there are enough blood vessels remaining intact.
In a case such as Gregg it is even clearer that the doctor treats the patient as though he has a
chance of recovering. When Mr Gregg was eventually referred to a specialist who diagnosed the
99
Beever, ‘Gregg v Scott’ (n39) 209.
Gregg (HL) (n26) [42].
100
150
lump as cancer he was told that he had a 10-15percent chance of survival if treated.101 Even
armed with knowledge of the extent and location of the cancer, and with understanding of
factors that affect survival rates, the doctor was unable to say whether Mr Gregg was one of the
10-15 out of 100 who would survive or one of the 85-90 out of 100 who would die even with
treatment.
Since this argument is based on the medical reality and shared experience of the doctor and
patient, it would apply in all cases of misdiagnosis/mistreatment of existing illness. Since it is not
an exceptional solution to a problem of proof of causation there would be no need to seek to
circumscribe its application by reference to vague or arbitrary criteria. In contrast, the loss of
chance argument for proportionate recovery would enable claimants to bypass the causation
requirement, so its supporters have sought to limit its application by reference, for example to
the degree of uncertainty in the disease. Jansen argues that ‘chance’ exists where there is ‘a
genuine, unavoidable uncertainty, relative, of course, to standards of scientific knowledge’ so in
his example the question of whether someone will survive a cardiac arrest ‘can only be answered
in probabilities and is therefore best perceived in terms of chance and risk’.102 Similarly Lord
Nicholls in Gregg suggested that a loss of chance approach to proportionate recovery was
appropriate there but not in Hotson because the patient’s prospects in Gregg but not in Hotson
were ‘fraught with a significant degree of medical uncertainty’.103 Lord Nicholls acknowledged
that this might be criticised for being ‘an imprecise boundary’,104 and such criticism is arguably
well-founded. Most cases of medical negligence will involve a degree of uncertainty surrounding
causation and that is precisely why the law only requires proof on the balance of probabilities. If
courts allowed proportionate recovery in some cases where they felt that the degree of
uncertainty was great enough, decisions would essentially turn on whether the court felt that the
Gregg (CA) (n28) [13] (Latham LJ). This was his ‘chance’ at the time of diagnosis – by time of trial his chance was
assessed at 20-30%.
102 Nils Jansen, ‘The Idea of a Lost Chance’ (1999) 19 OJLS 271, 279.
103 Gregg (HL) (n26) [49].
104 ibid.
101
151
claimant was deserving of help in avoiding difficulties of proof. In contrast, recognising that
every patient who is already ill has a chance of being cured, and that this chance is something of
value to him independently of the outcome (i.e. not leading to proportionate recovery for the
physical harm) because at the time of treatment both the patient and the doctor see the outcome
as being uncertain, then loss of chance would be recognised in every case where medical practice
is to describe the patient as having a chance. When the concept of chance is analysed on its own
terms as arising out of the considerations of interpersonal responsibility unique to the
doctor/patient relationship there is no need to try to draw lines to limit it to being an exceptional
approach.
4.2.3
When does a patient have a ‘chance’? Chance vs. risk.
It is still necessary to address the concern that loss of a chance is no different from an increase in
risk. It will be argued that this concern arises because recovery for loss of a chance is usually
presented as a solution to difficulties of proof of causation and as leading to proportionate
recovery. This bears some resemblance to the application of the Fairchild principle in Barker
which led to proportionate liability.105 By considering chance independently from the outcome,
and as arising because of unique characteristics of the doctor-patient relationship, it will be clear
that chance is conceptually distinct from risk. This is because it arises in the limited context of
medical misdiagnosis or mistreatment of existing illness where the shared uncertainty of the
doctor and the patient means that ‘chance’ becomes a meaningful concept.
Jones has noted the apparent similarities between chance and risk:
Although their Lordships would no doubt protest that the facts of Barker are at some
remove from Gregg v Scott, the acceptance that an increase in the risk of harm can
constitute actionable damage (where the harm has actually materialised) indicates that
there is no conceptual objection to a claim based on the loss of a chance of a better
Fairchild v Glenhaven Funeral Services (t/a GH Dovener & Son) [2002] UKHL 22, [2003] 1 AC 32; Barker v Corus
(UK) plc [2006] UKHL 20; [2006] 2 AC 572. These cases are addressed in Chapter Four.
105
152
medical outcome. The explanation for the different outcomes comes down to a choice as
to which claimants, faced with causal uncertainty, deserve to be given a helping hand in
the litigation lottery.106
If loss of a chance did lead to proportionate recovery then it would be conceptually close to risk.
The claimant in Barker suffered mesothelioma caused by asbestos inhalation but the ‘evidentiary
gap’ in relation to the aetiology of that disease means that where the asbestos has come from a
number of former employers it is impossible to say on the balance of probability that any
individual source was a cause of the harm. Exceptionally the Fairchild test of ‘material
contribution to the risk of harm’ applies. In Barker the House of Lords held that the Fairchild test
should lead to several liability so that each defendant is liable only for the extent of his
contribution to the overall risk. Given that the overall risk was a risk of the physical harm that
the claimant suffered, this amounted to liability for a portion of the physical harm. This was a
defendant-focused solution, the claimant was still notionally entitled to compensation for the full
value of the physical harm (subject to contributory negligence) but each defendant was only
liable for a portion of the total loss according to the extent of his contribution to the total risk of
harm. The loss of chance approach proposed by the claimants in Hotson and in Gregg awards
partial recovery for the claimant although the effect would be comparable for the defendant
doctor who would compensate a proportion of the physical harm corresponding to his
contribution to the overall risk.
Yet the risk-based approach was applied in Barker because of an evidentiary gap relating to the
overall causal process, whereas the problem in Hotson and in Gregg was that despite the
understanding of the causal process the claimant was unable to prove his individual case. In
terms of the NESS test, the claimants in Hotson and in Gregg were unable to prove that the
negligence was a necessary element of the sufficient set. With better evidence he would have been
able to prove that the negligence was a cause of his loss. But in Fairchild and Barker the general
106
Jones (n84) 268.
153
lack of understanding of the aetiology of the disease meant that one could never prove a causal
link. The uncertainty there was what constitutes a sufficient set in the first place let alone requiring
the claimant to prove that the negligence was necessary on this occasion.
The problems of proof are therefore very different in Barker and in the loss of chance cases, but
if proportionate recovery was allowed in loss of chance cases it would effectively achieve the
same outcome by holding the defendant liable for his contribution to the total risk. This would,
therefore, have extended the boundaries of the Fairchild exception. Limits to the application of
proportionate liability would have to be sought beyond corrective justice and in policy
arguments. For example, Green considers whether the Fairchild exception could be applied in a
case such as Chester v Afshar and beyond noting that Fairchild addressed the specific problem of
the evidentiary gap, she turns to policy arguments to distinguish the cases, noting that
Fairchild was concerned with a claimant who had been injured by his employer, whilst
that employer was acting solely in pursuit of commercial gain. This is not obviously
comparable to a claimant who has been injured during a surgeon’s attempt to ameliorate
a naturally occurring injury.107
In contrast, recognising that the chance of a cure is valued by the claimant and defendant
independently of the physical outcome, is consistent with corrective justice because it simply
recognises damage to this chance as a loss caused by the defendant. It does not lead to
proportionate recovery for the physical outcome so it is conceptually distinct from the notion of
contribution to risk.
4.2.3.1 When does a patient have a meaningful chance?
It is also important to note that a patient only has a meaningful chance when he is already unwell
and the negligence relates to the treatment of this illness. It is the existing illness that creates the
shared uncertainty for him and the doctor. In contrast, if the doctor introduces a new risk of
107
Green, ‘Coherence of medical negligence cases’ (n39) 19.
154
harm then the situation is like any other case of negligence and it is not meaningful to talk of the
claimant having a ‘chance’. This means that the approach to loss of chance suggested here would
not alter the decision in Wilsher v Essex Area Health Authority.108 In that case the claimant was a
baby who had been born prematurely and later suffered blindness caused by a condition called
retrolentalfibroplasia (‘RLF’). While in hospital he received negligent medical treatment when a
catheter was inserted into a vein rather than an artery leading to incorrect measurement of the
oxygen levels in his blood and the hospital therefore administering excess oxygen. Excessive
oxygen in the blood is known to cause RLF, but due to his being born prematurely the baby also
suffered four other conditions that each also carried a risk of RLF. He was unable to prove on
the balance of probabilities that it was the excess oxygen rather than one of the naturally
occurring conditions that caused his blindness and the House of Lords refused to apply the
McGhee test of material contribution to the risk of harm in this case.109 Since it was not possible
to identify on the balance of probabilities which harmful agent (e.g. oxygen) had caused the
illness, it was not possible to say that the defendant had contributed to the relevant risk factor.
Stapleton has argued:
[T]he pure loss-of-a-chance claim in Gregg might indirectly have put pressure on the
decision on Wilsher which has been recently and unanimously approved by the Lords in
Fairchild. At the moment after breach might baby Wilsher have had a loss of a chance
claim of the sort contemplated by Lord Nicholls because his premature condition gave
rise to a “significant medical uncertainty”? Even the adoption of the control mechanism,
or “artificial hook”, of requiring that the breach had caused a physical change might not
prevent such a pre-blindness claim being made by baby Wilsher, who could argue that
the excessively oxygenated blood resulting from breach was such an “injury”.110
[1988] AC 1074 (HL).
The McGhee test was applied in Fairchild and is more commonly known as the Fairchild test today, but Wilsher was
decided in the interval between those two decisions.
110 Stapleton, ‘Loss of the chance of cure’ (n10) 1003.
108
109
155
There are two issues to address here. The first relates to the ‘hook’ argument. This has already
been discussed in detail to show that there would have to be a causal relationship between the
physical ‘hook’ and the related chances. It is important to note that a physical change does not
constitute damage until it makes the claimant worse off, so excessively oxygenated blood would
not constitute a ‘hook’.111
The second issue is that in Wilsher the doctor exposed the baby to a discrete risk, independent of
the naturally occurring risks, so the baby should not be described as having a ‘chance’ of
avoiding RLF that the doctor caused him to lose. The fact that the claimant could not show that
it was oxygen that caused the harm meant that the House of Lords held that the defendant could
not be shown to have contributed to the relevant risk, so the negligence had not made a ‘material
contribution to the risk of harm’. Understanding this also explains why a patient may lose a
chance of cure when a doctor fails to diagnose or treat his existing illness, but does not lose a
chance when the doctor introduces a new risk.
Lunney suggests that Gregg would affect the analysis of Wilsher suggesting that ‘both Wilsher and
McGhee are examples of attempts to deal with evidentiary uncertainty as to causation and could
all be described as loss of chance cases. In Wilsher, it could be argued that the plaintiff lost the
chance that he would not have contracted RLF from non-negligent exposure to excess oxygen’.112
This is surprising because in the introduction to his article, Lunney says:
The exact definition of ‘loss of chance’ is unclear, but it is generally accepted that it refers
to the loss of an opportunity to obtain or receive a desired outcome; for example,
recovery from a personal injury. Accordingly, where a patient receives negligent
treatment in circumstances where, even if the treatment was properly carried out, that
patient might not have been cured it is said that the patient has lost the chance of being
cured as a result of the treatment.113
Rothwell v Chemical and Insulating Co Ltd [2007] UKHL 39, [2007] 3 WLR 876.
Mark Lunney, ‘What price a chance?’ (1995) 15 LS 1, 8.
113 ibid 1.
111
112
156
Although Lunney did not set this out as a definitive definition, the fact that he said that chance
arises in circumstances where, even if the treatment was properly carried out, that patient might not have been
cured implies that he sees chance as arising only where there is an existing risk and it is that risk
that the negligence allows to develop into harm. Stapleton has said that:
[I]n Wilsher at least one source of the risk was known to have been an innocent source
entirely unrelated to the defendant. If the loss of a chance claim had succeeded in Gregg,
also a case involving a known unrelated unequivocally innocent source, this would have
suggested that a moment after breach a future baby Wilsher would have a claim.114
But although the source of risk in Gregg was innocent it was not ‘entirely unrelated’. The cancer
arose independently of the doctor, but it was not unrelated to the doctor’s negligence because it
was this specific risk that the doctor was supposed to identify. In the approach to loss of chance
being proposed in this chapter it is clear that a patient has a ‘chance’ of recovery when he is
already unwell and the medical negligence consists of a failure to diagnose or treat that particular
illness meaning that the likelihood of the patient recovering from that illness decreases. The
concept of chance is meaningful in that context because the patient is already at risk, the doctor’s
duty is focused on that specific risk, and both the doctor and the patient are faced with
uncertainty regarding the patient’s prognosis.
If the doctor introduces the patient to a new risk through negligent mistreatment, such as adding
excessive oxygen to the patient’s blood, then he might cause physical harm but he has not
damaged the patient’s ‘chance’ of recovering because the patient did not have a meaningful
‘chance’ of avoiding that harm. To say that the patient has a chance of avoiding new illnesses is
too general a conception of chance for it to be meaningful. In contrast it is meaningful to
describe a patient who is already unwell as having a chance of recovering from that illness
because his prospects are already damaged and uncertain.
114
Stapleton, ‘Loss of the chance of cure’ (n10) 1003.
157
A chance exists because the patient’s illness creates an uncertainty that is specific to him rather
than the general level of uncertainty to which background risks expose everyone. For example, in
hospital there is a general background risk of contracting MRSA and every patient is exposed to
this risk so it is not meaningful to describe an individual as having a ‘chance’ of avoiding it. But if
a patient is already suffering from MRSA and the doctor negligently failed to diagnose it then the
likelihood of that individual dying from it was already higher than the normal person and his
prognosis was uncertain so he had a meaningful ‘chance’ of avoiding the harmful outcome.
4.2.4
‘Loss’ of a chance
It is important to remember that there must be a ‘loss’ of a chance and that the doctor’s
negligence must be shown, on the balance of probabilities, to be a cause of this loss.
When loss of chance is presented as a solution to the difficulty of proving causation, one
objection is that it is not possible to say that the chance has been ‘lost’ unless it has been reduced
to zero. In other words, it is only once the claimant has suffered physical harm that his chance
has been definitively lost. In Gregg, Lord Phillips therefore left open the question of whether
proportionate recovery for loss of a chance would be appropriate where physical harm had
actually occurred, but on the facts of Gregg he considered it impossible to say whether a chance
had been lost because the physical harm had not yet occurred. This problem simply does not
arise in the approach proposed in this chapter where the concept of chance is meaningful
because at the time of the diagnosis and treatment of the illness both the doctor and the patient
regard the prognosis as being uncertain and therefore value the statistical assessment of the
patient’s chance of cure. This means that it is the fact of a reduction in the statistical chance that
constitutes a ‘loss’ of a chance.
Lunney has argued that in the medical negligence context it is inappropriate to talk of ‘loss of a
chance’ because the chance is not definitively lost. The patient still undergoes treatment and the
remaining chance runs its course. For him this differs from cases such as Chaplin where the
158
claimant was definitively deprived of the opportunity to enter the beauty contest. Lunney
explains this through the analogy of a racehorse injured by a driver’s negligence on its way to a
race. He says that if the accident prevents the horse from racing, then the negligence has
‘deprived the horse of the chance to win the race [so] this would be a true lost chance case’. If,
however, the horse suffers shock but is still able to run and finishes seventh there would be no
claim for the loss of the chance of a better outcome: ‘the horse has run the race and has lost, and
either the accident caused the horse to lose or it did not. There is no question of chance; it is for
the court to decide whether the negligence was responsible for the outcome (the horse losing the
race)’.115
However arguably it is the fact that in the medical negligence context the patient has to run the
chance with decreased odds that makes the chance so significant and valuable. In the context of
a race, the horse stands to make a gain. In the medical negligence context the patient stands to
suffer illness and possibly death. The owner of the horse has the choice whether or not to run
the race with lower odds. The patient has no choice but to undergo treatment with a lower
chance of recovery. Furthermore, the risk of the horse losing the race is a risk that is
independent of the driver. In contrast, where a patient is already ill they have a chance of
recovering and it is the purpose of the defendant doctor’s duty of care to diagnose and treat that
illness and protect the patient’s chance of recovery.
Traditional negligence principles would still apply to this damage, so the claimant must prove on
the balance of probabilities that the negligence caused the reduction in his chance of cure. In
Hotson the judge found that the delay had allowed the bleeding to continue to build pressure in
the joint and block those blood vessels that had remained intact, turning his injury into an
inevitability. This meant that his physical state had worsened so his chance of cure was reduced
by the delay caused by the negligence. In Gregg the claimant’s cancer spread during the nine
months delay so it was clear that his physical state had worsened and this was responsible for the
115
Lunney (n112) 5.
159
reduction in the statistical likelihood of cure. If the cancer had remained unchanged during that
time then it would be hard to see that the delay caused any reduction in the chance of cure.
4.2.5
Valuing the chance
The final question is how to value this chance. In this approach the lost chance exists
independently of the physical outcome and this has important consequences for the valuation of
the chance because it means that the valuation must also be independent of the final outcome. It
also means that where a chance has been lost, it is actionable regardless of whether or not the
illness was cured. If a claimant suffers the physical harm and has a successful claim in respect of
that outcome he should still be able to recover for the loss of a chance because that was valuable
to him during his treatment – it is a distinct head of damage.
In contrast, the claims that have been brought on the basis of loss of a chance have sought to
recover a proportion of the value of the physical injury. In Hotson the trial judge found that the
claimant had lost a 25 percent chance of avoiding avascular necrosis and awarded 25 percent of
the value of this physical damage. In Gregg, the claimant argued that his chance of survival had
been reduced from 42 percent to 25 percent and the claim was for a proportion of the value of
the loss that he would suffer if he failed to survive i.e. a proportion of the value of the physical
injury itself and of the loss of lifetime’s earnings.
Hogg has argued that proportionate recovery is a fair solution:
[L]ost chance recovery provides an equitable via media to the problem of causal
uncertainty…Such an approach strikes a balance between letting off a defender who may
in fact have caused the harm, and penalising a defender who may not have caused the
harm at all.116
This justification is based on economic efficiency or on vague notions of fairness and is
incompatible with corrective justice based liability. Voyiakis explains:
116
Martin Hogg, ‘Re-establishing Orthodoxy in the Realm of Causation’ (2007) Edin LR 8, 15.
160
Knowing how much risk I have imposed on you and how much compensation I would
have to pay you if I had caused you actual physical harm does not by itself suggest how
much I should pay for having exposed your physical health to danger. To take a crude
example, when I make it 40 per cent more likely that you will suffer lung cancer, I am not
causing you 40 per cent of the harm that lung cancer brings about. The quantity of risk
and the quantity of physical harm do not seem to be related in any such straightforward
way. We therefore have reason to object to the latter being taken as a measure of the
former.117
This much is apparent if we recall the more straightforward causal models addressed in Chapter
Two. There it was said that if the harm is indivisible then each of the necessary elements of the
sufficient set is a cause and each is a cause of the whole loss. Apportionment of liability might be
appropriate as a matter of moral responsibility, but each cause has causal responsibility for the
whole of the harm. In contrast, where the harm is divisible i.e. dose-related, then each causal
factor contributes/causes a portion of the overall harm. Thus a defendant who has contributed
40 per cent of the harmful agent has caused 40 per cent of the harm. So it seems illogical to say
that a doctor who has contributed a 40 percent risk to an indivisible harm should pay 40 percent
of the value of the physical harm. Valuing the chance as a percentage of the physical harm
reinforces the sense that the loss of a chance argument as it was advanced in Hotson and in Gregg
is simply an attempt to assist a claimant to bypass the causation requirement.
The argument made here, instead, is that the loss of a chance is an actionable loss independent
of any physical harm. It should, therefore, be treated as any other loss and assigned a value
within the Judicial College’s Guidelines for the Assessment of General Damages in Personal Injury Cases.118
Jansen has addressed the ‘pragmatic objection [which] emphasizes the difficulties of assessing the
value of chances’:
Emmanuel Voyiakis, ‘The Great Illusion: Tort Law and Exposure to Danger of Physical Harm’ (2009) 72 MLR
909, 917.
118 (n83).
117
161
[T]here are difficulties in assigning a financial value to lost chances, especially if they are
related to physical harm…It is difficult to evaluate physical injuries financially. The
widely differing amounts which courts award to physically injured parties illustrate that.
But this indicates only that there is no ‘objective’ measure as such, and that the measure,
therefore is to be determined by policy considerations. Courts do manage to deal with
this consistently.119
In other words, the value assigned to any physical injury is necessarily artificial. This has been
addressed by using the tariff system so that the level of damages is commensurable with the
seriousness of the injury. Likewise, in relation to risk, Voyiakis argues that while the value of the
risk is not necessarily an equivalent percentage of the physical harm, we have an intuitive notion
that some risks are more serious than others. He explains:
This intuition seems to follow naturally from the idea that protecting oneself against risk
of some harm is a means to protecting oneself against the actual harm. It also seems to
tally with our intuitive comparisons across risks of different seriousness: the effect of
being exposed to a 40 per cent risk of lung cancer is more serious than the effect of
being exposed to a 40 per cent risk of a broken toe exactly because getting lung cancer is
much more serious than getting a broken toe.120
This means that it would be possible to set a tariff for a lost chance. Any physical injury is
represented by a range of values in the tariff, with the judge being able to use the range of the
scale to reflect the impact the injury had on the individual. ‘Loss of a chance’ could have a similar
scale. It would be possible to set a value for the loss of a chance of avoiding a life-threatening
illness with a scale of damages to reflect a small reduction in chance or a large reduction in
chance, another value for the loss of the chance of avoiding a less serious illness (or disability or
temporary illness) with a similar scale to reflect the extent of the reduction in the chance.
119
120
Jansen (n102) 293.
Voyiakis (n117) 917.
162
Conclusion
In summary, the loss of chance claim for proportionate recovery is rejected in this thesis because
a patient does not have an objective chance of being cured. This means that proportionate
recovery effectively awards damages for the physical harm with a discount to reflect the degree
of doubt surrounding proof of causation. Instead this thesis proposes a more limited role for
loss of a chance as an independent form of damage leading to limited recovery. Such loss only
arises in a meaningful way because of the special nature of the doctor-patient relationship, and
only in relation to misdiagnosis/mistreatment of existing illness.
163
Chapter Four: The evidentiary gap
Introduction
This chapter now addresses the problem of the ‘evidentiary gap’ that arose in McGhee1 and in
Fairchild.2 In this chapter it is necessary first to analyse the relevant case law in order to
understand what exactly the so-called ‘evidentiary gap’ is. The first section will draw on the
NESS analysis of causation presented in chapter two to locate and fully understand the problem
that arose in these cases. Whereas claimants more commonly face the problem of proving that
the negligence was a necessary element of a sufficient set of conditions that actually occurred, it
will be shown that where there is an ‘evidentiary gap’ the claimant faces the prior problem that
the state of scientific uncertainty prevents the definition of a sufficient set of conditions for the
harm that he has incurred. The subsequent sections will turn to the legal solutions that have been
adopted to address the evidentiary gap. Beginning with McGhee, and continuing through Fairchild,
and Barker,3 the courts have considered it sufficient for the claimant to prove that the
defendant’s negligence materially increased the risk of the relevant injury, or materially
contributed to the risk of the relevant harm, in cases involving an evidentiary gap.
One interpretation of this test is that it simply alters the legal standard of proof that the claimant
is required to meet, so section two will consider whether this is an appropriate solution. Related
aspects of proof will also be considered, such as the place of statistics and the possibility of
proof of causation on traditional principles through proof that the negligence more than doubled
the background risk. Ultimately it will be argued that given the gap in scientific understanding
surrounding the aetiology (the ‘evidentiary gap’) it is inappropriate for the law to attempt to
bridge this gap – such resort to intuition in the face of explicit scientific uncertainty would base
McGhee v National Coal Board [1973] 1 WLR 1 (HL (Sc.)).
Fairchild v Glenhaven Funeral Services (t/a GH Dovener & Son) [2002] UKHL 22, [2003] 1 AC 32.
3 Barker v Corus [2006] UKHL 20, 2 WLR 1027.
1
2
164
liability on a fiction. The result is that the defendant is liable even though his negligence may not
have been a cause of the harm.
Section three will therefore turn to the alternative interpretation of the Fairchild test, as proposed
in Barker, that in cases where the test applies the damage which forms the gist of the negligence
action is the increase in risk.4 Liability under Barker was several (rather than joint and several) and
the extent of an individual defendant’s liability was to be measured by the extent of his
contribution to the total risk to which the claimant was exposed. At first glance this solution
appears to conform to corrective justice because the defendant is held liable only for that loss
which he has been proved to have caused i.e. since he has only been proved to have increased
the risk of the claimant’s injury, he is only held liable for this increase in risk. However, closer
analysis shows a number of flaws in this approach. By insisting that the claimant have suffered
the relevant physical harm before allowing him to bring a negligence action for exposure to the
risk of this harm the effect is that the physical harm remains the gist of the action and the
apportionment of liability that takes place reflects the possibility that the defendant caused the
harm. (This echoes the criticism of proportionate recovery for ‘loss of a chance’ – the reality is
that it is liability for possible causation of harm.) Furthermore, liability for risk is a more deeply
problematic concept. Risk is a forward-looking concept so is mismatched with the backwardlooking causation enquiry. More significantly, risk lacks moral significance as damage in the
context of the negligence enquiry because it does not add anything to the breach of duty
requirement. Conduct is wrongful if the defendant failed to take reasonable care i.e. he exposed
the claimant to an unreasonable risk of harm. If ‘exposure to risk’ also constitutes actionable
damage then the damage requirement would effectively be subsumed into the breach inquiry.
Although the majority in BAI v Durham [2012] UKSC 14, [2012] 1 WLR 867 considered that the gist of the action
was still the physical harm and that the decision in Barker had simply equated ‘materially increasing the risk’ with
‘contributing to the cause’, Lord Phillips (dissenting) insisted that the rule in Barker was that where the Fairchild test
of ‘material increase in the risk of harm’ is applied the gist of the action is the risk of harm.
4
165
This is inconsistent with corrective justice which is concerned with ‘wrongful loss’ not with
wrongdoing in isolation.5
The final section will refocus on corrective justice and argue that these claims ought to fail for
want of proof of causation since the approaches explored in previous sections fail to achieve
consistency with corrective justice. Since the defendant is liable in circumstances where he may
not have been a cause of the loss, corrective justice is abandoned in these cases in favour of the
pursuit of consequentialist goals. The effect of this has been to throw negligence into the state of
incoherence that is inherent in the pursuit of consequentialist goals within the bipolar framework
of the tort action. Barker seems to achieve a fair balance between the competing interests of
claimants and defendants, but this is simply the author’s personal preference since there is no
logical demand on the limits of liability. Ultimately it is preferable that the scope of application of
the Fairchild test be as tightly circumscribed as possible so that the incoherence that has affected
liability in these cases is not repeated in other contexts.
1. Defining the evidentiary gap
The so-called ‘evidentiary gap’ problem was first addressed in McGhee v National Coal Board 6 in
relation to the cause of dermatitis, and subsequently in relation to the cause of mesothelioma in
Fairchild v Glenhaven Funeral Services (and later cases).7 The defendant in McGhee was held to be
liable for the claimant’s dermatitis on the basis that his negligence had materially increased the
risk of dermatitis. Following this decision it was unclear whether this constituted an exceptional
new ‘test’ for causation or whether the court had simply taken a ‘robust’ view of the evidence
Porat and Stein have also suggested that the gist of the action in evidentiary gap cases could be reconceptualised as
‘evidentiary damage’. They suggest that if there was only one source of the harmful agent the claimant would not
face any problem of proof, so each defendant who exposes the claimant to another source of the harmful agent is
responsible for preventing the claimant from having a successful negligence action. This argument has not been
widely accepted, and raises a distinct question as to whether a claimant has a right to protect the integrity of his
evidence from indirect interference. The evaluation of this right falls outside the scope of this thesis. See Ariel Porat
and Alex Stein, Tort Liability Under Uncertainty (Oxford University Press 2001).
6 McGhee v National Coal Board 1972 SLT (notes) 61 (Court of Session, Scotland); [1973] 1 WLR 1 (HL (Sc.)).
7 Fairchild v Glenhaven Funeral Services (t/a GH Dovener & Son) [2001] EWCA Civ 1881, [2002] 1 WLR 1052; [2002]
UKHL 22, [2003] 1 AC 32.
5
166
before it. The House of Lords held that the test did not apply to the facts of Wilsher v Essex
AHA,8 but did apply in Fairchild. The purpose of this section is to examine the causal problem
that arose in each of these cases, and the scientific evidence that was available, in order to
understand precisely what the ‘evidentiary gap’ consists of. Approaching this from the basis of
the NESS test will be shown to have a dual advantage over the but-for test. It first enables the
causal problem to be conceptualised more accurately. This then means that the possible legal
responses to the problem of the evidentiary gap can be analysed with greater clarity.
1.1 The evidentiary gap relating to dermatitis: McGhee v National Coal Board
The claimant in McGhee developed dermatitis as a result of damage to his skin from brick dust.
He became covered in dust while emptying brick kilns for the defendant employer. This
exposure to brick dust at work was unavoidable, so it was not negligent. The defendant’s
negligence consisted of failure to provide shower facilities to enable the claimant to wash the
brick dust off his skin before cycling home. The medical evidence was that the claimant’s
dermatitis was caused by brick dust on his skin, so both the innocent and guilty periods of
exposure to brick dust were potential causes of the dermatitis in this case. The problem facing
the claimant was to prove that the negligent exposure in particular was an actual cause of his
dermatitis. On the basis of the available legal tests he was required to prove that but-for the
negligent exposure he would not have developed dermatitis, or, applying the Wardlaw test,9 that
the negligent exposure had materially contributed to his dermatitis.
The difficulty the claimant faced in this particular case was that although medical science was
able to state that brick dust does cause dermatitis, it was unable to explain how it causes
dermatitis.10 This meant that proof that the negligent exposure had actually contributed to the
claimant’s disease was problematic. Lord Wilberforce noted that the experts ‘had to admit that
Wilsher v Essex AHA [1988] AC 1074 (HL).
Bonnington Castings v Wardlaw [1956] AC 613 (HL).
10 McGhee v National Coal Board [1973] 1 WLR 1 (HL) 3 (Lord Reid).
8
9
167
they knew little of the quantity of dust or the time of exposure necessary to cause a critical
change’.11 Lord Reid explained this gap in the medical experts’ understanding more clearly:
In the present case the evidence does not show – perhaps no one knows – just how
dermatitis of this type begins. It suggests to me that there are two possible ways. It may
be that an accumulation of minor abrasions of the horny layer of the skin is a necessary
precondition for the onset of the disease. Or it may be that the disease starts at one
particular abrasion and then spreads, so that multiplication of abrasions merely increases
the number of places where the disease can start and in that way increases the risk of its
occurrence.12
The ‘evidentiary gap’ in this case therefore concerns the aetiology of the disease, in particular the
question of whether it is caused by an accumulation of abrasions or by a single abrasion. If it
were caused by a single abrasion then the but-for test would apply, and if it were caused by an
accumulation of abrasions the test of ‘material contribution to harm’ would apply. Since the
causal process is unknown, neither test can be satisfied.
1.2 The evidentiary gap relating to mesothelioma: Fairchild v Glenhaven Funeral
Services
The victim in Fairchild had died from mesothelioma, a malignant tumour in the pleura (the
membrane surrounding the lungs) and the negligence action was brought by his widow.13 During
his working life he had been exposed to asbestos by a number of his former employers, one of
whom was the defendant employer. The problem facing the claimant was to establish that the
defendant’s negligence was a cause of the victim’s disease. Similarly to McGhee it was known that
inhalation of asbestos fibres does cause mesothelioma, but medical science was unable to explain
how it causes mesothelioma. Once again, therefore, the problem of proof arose because of the
ibid 5 (Lord Wilberforce).
ibid 4 (Lord Reid).
13 Fairchild v Glenhaven Funeral Services Ltd and others [2002] UKHL 22, [2003] 1 AC 32, [7].
11
12
168
scientific uncertainty combined with the fact that there was more than one period of exposure to
the harmful substance.
Lord Bingham explained the uncertainty:
The mechanism by which a normal mesothelial cell is transformed into a mesothelioma
cell is not known. It is believed by the best medical opinion to involve a multi-stage
process, in which six or seven genetic changes occur in a normal cell to render it
malignant. Asbestos acts in at least one of those stages and may (but this is uncertain) act
in more than one…It is accepted that the risk of developing a mesothelioma increases in
proportion to the quantity of asbestos dust and fibres inhaled: the greater the quantity of
dust and fibre inhaled, the greater the risk. But the condition may be caused by a single
fibre, or a few fibres, or many fibres: medical opinion holds none of these possibilities to
be more probable than any other, and the condition once caused is not aggravated by
further exposure.14
The scientific uncertainty concerning the quantity of asbestos fibres required to cause
mesothelioma, and the stage(s) at which the fibres are involved in the causal process, combined
with the fact that the claimant had been exposed to asbestos by a number of former employers,
meant that he was unable to establish causation, either on the but-for test or on the ‘material
contribution to harm’ test because it was impossible to say what he needed to prove. Stapleton
has explained:
This means that there is as yet no means of telling whether the mesothelioma of a person
subjected to a sequence of asbestos exposures was due to all exposures, only some or
one, let alone which one. Importantly, there is no direct basis for saying that longer,
more intense exposures are more likely to have been the cause…than much shorter
14
ibid (Lord Bingham).
169
exposures, nor is there any basis for saying that earlier exposures are more likely to have
been the cause than later exposures.15
Lord Bingham called this a ‘rock of uncertainty’, and it is generally known as the ‘evidentiary
gap’.16 The evidentiary gap is therefore a very specific evidential problem. It is not simply a
problem of a lack of evidence relating to the individual claimant. It goes to the general state of
knowledge and understanding of the disease in medical science. The ‘evidentiary gap’ must also
be described more accurately than ‘scientific uncertainty about the aetiology of the disease’. The
disease is known to be indivisible (i.e. its severity is not dose-related), and the harmful agent is
known (e.g. brick dust, asbestos), but the process by which it is caused may be an accumulation or
interaction of a few or many fibres.17
Little has changed in the understanding of the aetiology of mesothelioma in the years since
Fairchild, as Lord Phillips explained in the recent case of Sienkiewicz v Greif.18 It is still believed
that a small number of cases may possibly be idiopathic, meaning they have an unknown cause
other than asbestos, but mesothelioma is ‘always, or almost always, caused by the inhalation of
asbestos fibres’.19 One development since the decision in Fairchild concerns the minimum length
of the latency period which is now believed to be five years rather than ten. This period is
important because any exposure to asbestos during this period has no causative effect because
the disease has already been caused. Furthermore, it is no longer believed that mesothelioma is
Jane Stapleton ‘Lords a’leaping evidentiary gaps’ (2002) 10 Torts LJ 276, 281.
See for example Stapleton ‘Lords a’leaping’ (n15).
17 It is therefore distinct from the problem that arose in Wilsher. The claimant there was a baby born prematurely
who developed retrolentalfibroplasia (‘RLF’). The defendant had negligently exposed the claimant to excess oxygen
which is known to cause RLF, but the claimant also suffered a number of other conditions that were equally capable
of causing RLF and which were due to his premature birth rather than to negligence. The case will be examined in
greater detail at a later stage, but it is important to note that the problem in that case was not that medical science
was unable to explain how RLF is caused, but that the claimant was unable to show that excess oxygen was a cause
in his individual case.
18 Sienkiewicz v Greif (UK) Ltd [2011] UKSC 10, [2011] 2 AC 229. Lord Phillips based this understanding on a case
control study by Peto and Rake, published in 2009 by the Health and Safety Executive, on “Occupational, Domestic
and Environmental Mesothelioma risks in Britain” (“the Peto Report”), which he explained ‘is said to be the first
representative study to quantify the relationship between mesothelioma risk and lifetime occupational and residential
history in this country’ (at [18]), and on ‘the collation of data about mesothelioma set out by Rix LJ in his judgment
in the series of appeals collectively described as Employers’ Liability Insurance “Trigger” Litigation [2010] EWCA Civ
1096’ (at [19]).
19 Sienkiewicz (n18) [19] (Lord Phillips).
15
16
170
triggered by a single asbestos fibre.20 While it is therefore believed that mesothelioma is caused
by multiple asbestos fibres, there remains significant scientific uncertainty as to how it is caused
and at what stages the fibres are effective. Lord Phillips explained:
It is believed that a cell has to go through 6 or 7 genetic mutations before it becomes
malignant, and asbestos fibres may have causative effect on each of these.
It is also possible that asbestos fibres have a causative effect by inhibiting the activity of
natural killer cells that would otherwise destroy a mutating cell before it reaches the stage
of becoming malignant.21
He also noted another possible causal mechanism, explaining that ‘The Peto Report also raised
the possibility (but no more) of synergistic interaction between early and later exposures.
Causation may involve a cumulative effect with later exposure contributing to causation initiated
by an earlier exposure’.22
1.3 Analysis of the evidentiary gap
The focus of the chapter has so far been confined to determining the nature of the evidentiary
gap. The solution that was adopted in McGhee and applied in Fairchild was to hold the defendant
liable on the basis that his negligence had materially increased the risk of the harm suffered.
Before it is possible to engage in an analysis of this solution, and of the subsequent
developments to it in later cases, it is essential to understand more precisely the causal problem
that arises in these cases.
Steel and Ibbetson have explained that, ‘[t]he normal rule of causation…has two aspects: the
evidential and the conceptual. In order to establish liability, P has to show on balance of
probabilities (evidential) that but for D’s wrongful conduct the injury would not have occurred
ibid [102] (Lord Phillips).
ibid [19] (Lord Phillips).
22 ibid [102] (Lord Phillips).
20
21
171
(conceptual)’.23 Problems of causation divide into these two categories, the conceptual and the
evidential. Stapleton has explained that the problem in evidentiary gap cases is an evidential one:
‘[the claimant’s] difficulty was the antecedent one that there was simply inadequate factual
information to feed into the but-for test’.24 She therefore considers it irrelevant to address the
conceptual inadequacies of the but-for test because ‘this was not the problem facing the
claimants in Fairchild’.25 In this section it will be argued, however, that there are a number of
advantages to adopting the NESS test which was shown in chapter two to be conceptually more
robust than the but-for test. First, since the NESS test is better matched to the underlying
concept of causation it enables us to pinpoint precisely which aspect of causation the evidential
problem relates to. Second, the NESS analysis highlights the fact that the Wardlaw test of
material contribution to harm is distinct in nature from the McGhee/Fairchild test of material
contribution to the risk of harm. The former addresses a conceptual weakness in the but-for test,
so does not depart from the causation requirement, the latter addresses an evidential obstacle to
proof of causation regardless of how causation is conceptualised so represents a significant
exception to the causation requirement. Adopting the NESS test would not solve the evidential
problem raised by the evidentiary gap, but it would make it easier to understand exactly what is
at stake in these cases.
1.3.1 NESS and identifying the problem: locating the evidentiary gap
In Wright’s NESS account of causation, ‘a particular condition was a cause of (condition
contributing to) a specific consequence if and only if it was a necessary element of a set of
antecedent actual conditions that was sufficient for the occurrence of the consequence’.26
The problem that a claimant usually faces in negligence is to establish the necessity of the
defendant’s negligence in order to complete the sufficient set of conditions. An example is the
Sandy Steel and David Ibbetson, ‘More grief on uncertain causation in tort’ (2011) 70 CLJ 451, 452.
Stapleton, ‘Lords a’leaping’ (n15) 280.
25 ibid.
26 Richard Wright, ‘Causation in Tort Law’ (1985) 73 Cal L Rev 1735, 1790.
23
24
172
so-called ‘single hit hunters’ scenario that arose in the Canadian case of Cook v Lewis.27 If two
hunters have each negligently fired their guns in the victim’s direction and the victim has been hit
by only one bullet then the problem he faces is proving which hunter fired the bullet that hit him
i.e. which bullet was necessary for the sufficient set of conditions? Additionally the cases of Hotson
v East Berkshire HA and Gregg v Scott that were examined in the previous chapter involved an
evidential problem concerning the necessity of the negligence, and the claimants attempted to
avoid this problem of proof by proposing ‘loss of a chance’ as an alternative form of damage. It
was clear in these cases what they were required to prove, the difficulty was in proving it to the
balance of probabilities standard of proof. For example, in Hotson the sufficient set of conditions
for avascular necrosis is ‘insufficient intact blood vessels to keep the epiphysis alive’. The
problem for the claimant was to prove whether those blood vessels that were damaged during
the delay caused by the defendant’s negligence were necessary for this set, or whether they had
been pre-empted from having any effect by the quantity of blood vessels damaged during the
fall.28
In contrast, the claimants in the evidentiary gap cases of McGhee and Fairchild faced the prior
problem of identifying what constitutes a sufficient set of conditions to cause the relevant disease.
This was essential to their claims because there were multiple exposures to the harmful agent so
there were multiple potential causes. If medical science is unable to explain what constitutes a
sufficient set of conditions, then a claimant faces an insurmountable challenge – how can he
prove that the defendant’s negligence was necessary for the sufficiency of the set of conditions
to cause his dermatitis or mesothelioma if uncertainty surrounding the aetiology of his disease
means that medical science is unable to tell him what a sufficient set consists of? In other words,
in most negligence cases we know what the claimant needs to prove and the problem for him is
proving it to the requisite legal standard of proof. In cases involving this kind of evidentiary gap
27
28
[1951] SCR 830
See Ch 3 section 3.1.2.
173
it is not possible to say what the claimant needs to prove let alone asking him to prove it to a
particular standard.
The NESS test therefore enables the distinction to be made clearly between evidence of
causation in the personal sense of proving that the negligence was necessary for the sufficiency of
the set of conditions, and evidence of causation in the deeper, general sense of proving what a
sufficient set of conditions is for a given disease. The next section builds on this NESS-informed
understanding of the evidentiary gap in order to explore possible legal solutions with greater
clarity.
1.3.2 NESS: separating the conceptual from the evidential aspects of the legal
solution
The solution adopted by the majority of the House of Lords in McGhee built on the Wardlaw test
of material contribution to harm and suggested that there is ‘no substantial difference between
saying that what the defender did materially increased the risk of injury to the pursuer and saying
that what the defender did made a material contribution to the injury’.29 It will be argued that
there is, however, a significant difference between these two propositions since they are different
in nature. The Wardlaw test addresses a conceptual inadequacy of the but-for test whereas
McGhee addresses the evidentiary gap so it assists a claimant who faces an evidential barrier to
proof on any test. Therefore, McGhee is not simply an extension of the Wardlaw test, it is an
exception to it.
The problem with the decision in McGhee is that it seems to be built upon the misconceived idea
that the Wardlaw test of ‘material contribution to harm’ is an exceptional approach which
facilitates proof of causation and that making a further allowance where the claimant can only
prove a material increase in the risk of harm is not a significant leap. This is partly due to the fact
that, as chapter two illustrated, the speeches in Wardlaw did not adequately define the causal or
29
McGhee (n10) 5 (Lord Reid).
174
evidential problems in that case so the scope and effects of the test of material contribution to
harm are unclear. But it is also partly because the analysis in McGhee failed to define accurately
the causal and evidential problems raised by either case.
In McGhee, Lord Salmon and Lord Reid both observed that a distinction could be drawn between
the causal processes in Wardlaw and McGhee yet they chose to dismiss it as irrelevant. Lord Reid
explained that dermatitis may be caused by an accumulation of abrasions, or by one particular
abrasion, but continued to say:
I am inclined to think that the evidence points to the former view. But in a field where so
little appears to be known with certainty I could not say that that is proved. If it were,
then this case would be indistinguishable from Wardlaw’s case.30
He explained:
Nor can I accept the distinction…between materially increasing the risk that the disease
will occur and making a material contribution to its occurrence.
There may be some logical ground for such a distinction where our knowledge of all the
material factors is complete. But it has often been said that the legal concept of causation
is not based on logic or philosophy. It is based on the practical way in which the ordinary
man’s mind works in the everyday affairs of life. From a broad and practical viewpoint I
can see no substantial difference between saying that what the defender did materially
increased the risk of injury to the pursuer and saying that what the defender did made a
material contribution to his injury.31
This highlights that the ordinary man’s analogy between increasing the risk of harm and
contributing to the harm is flawed because he had previously explicitly stated that causation of
dermatitis is not analogous with the pneumoconiosis in Wardlaw. The danger of regarding
Wardlaw as taking an exceptional approach is that courts will not analyse the conceptual and
evidential problems that necessitated the ‘material contribution to harm’ approach, and will
30
31
ibid 4 (Lord Reid).
ibid 5 (Lord Reid). See also 12-13 (Lord Salmon).
175
instead focus on contextual similarities and policy considerations. Although Lord Wilberforce
adopted a different solution in McGhee, reversing the onus of proof, he still drew on contextual
similarities to Wardlaw in order to justify taking an exceptional approach. He focused on the fact
that both cases involved industrial disease and explained ‘[i]n cases concerned with
pneumoconiosis, the courts faced with a similar, though not identical, evidential gap, have
bridged it by having regard to the risk situation of the pursuer’.32 Likewise, in the later decision in
Fairchild, Lord Nicholls held that ‘[o]n occasions the threshold “but for” test of causal
connection may be over-exclusionary. Where justice so requires, the threshold itself may be
lowered’.33
It is essential to understand why the but-for test is over-exclusionary in order to assess whether
justice requires an exceptional approach. In the case of Wardlaw, the exception to the but-for test
was necessary because the but-for test was conceptually inadequate as a test of causation. In
McGhee and Fairchild the problem is not the conceptual aspect of the test for causation, but the
deeper evidential problem that it is not possible to define a sufficient set. As already stated,
adopting the NESS test would not overcome the problems facing the claimant because the
problem is an evidential one. The benefit of the NESS test is that is shows us that it is only
possible to recognise liability in these cases if an exceptional approach is adopted.
Now that the problem of the evidentiary gap has been explained, the focus of the chapter can
turn to the question of how to address this problem in a way that is consistent with corrective
justice. Sections Two and Three will consider a range of legal and academic responses in terms
of the concepts involved, before section Four returns the focus squarely on the demands of
corrective justice. Solutions fall into two broad categories: those that focus on the legal standard
of proof and those that seek to avoid the problem by reformulating the damage that forms the
gist of the negligence action. Section two is therefore focused on proposals that relate to the
32
33
ibid 6 (Lord Wilberforce).
Fairchild (n13) [40] (Lord Nicholls).
176
evidential standard, and section three considers attempts to reformulate the damage in these
cases.
2. Evidential solutions: the inferential interpretation of McGhee
The majority in McGhee found that the defendant had materially increased the risk of dermatitis,
and considered this a sufficient basis for liability. Lord Kilbrandon did not resort to reasoning
based on a material increase in risk, simply finding that causation had been established on the
balance of probabilities. Lord Wilberforce, however, pointed to the contradiction that lay at the
heart of the majority approach. He had made a similar assessment of the facts of the case,
explaining that pneumoconiosis involves a ‘similar, though not identical, evidential gap’.34 Yet he
preferred to resolve the problem by reversing the onus of proof because ‘to bridge the evidential
gap by inference seems to me something of a fiction, since it was precisely this inference which
the medical expert declined to make’.35
There were two competing interpretations of the significance of the finding of a ‘material
increase in risk’ following McGhee. It was initially held in Wilsher that the majority had simply
taken a ‘robust and pragmatic’ view of the available evidence in order to draw in inference of
causation.36 It was eventually held in Fairchild that this was an exceptional test for causation that
was applicable in tightly circumscribed cases to enable claimants to overcome the evidentiary
gap.37 However, in his corrective justice-based account of the tort of negligence, Rediscovering the
Law of Negligence, Beever argues that the majority in McGhee did draw a factual inference of
causation and that they were entitled to find that the claimant had succeeded in proving the
causal link on the balance of probabilities.38 Since the main arguments about causation in this
thesis are premised on a corrective justice account of negligence it is essential to examine the
validity of Beever’s assertions.
ibid 6 (Lord Wilberforce).
ibid 7 (Lord Wilberforce).
36 Wilsher (n8) 1090 (Lord Bridge).
37 Fairchild (n13) [22] (Lord Bingham), [44] (Lord Nicholls), [65] (Lord Hoffmann), [142] (Lord Rodger).
38 Allan Beever, Rediscovering the Law of Negligence (Hart Publishing 2007) 465-472.
34
35
177
It will be argued that the nature of the evidentiary gap means that it is simply not possible to
establish on the balance of probabilities that the defendant’s negligence was a cause of the
claimant’s disease. It may be the case that the scientific community generally requires a higher
standard of proof than the legal standard of the balance of probabilities, so there will be cases
where the expert evidence does not prove causation definitively yet a court can still justifiably
conclude that causation has been established on the balance of probabilities. But where the
expert evidence demonstrates an evidentiary gap there is simply insufficient information to feed
in to any test or in to the balance of probabilities standard of proof – no explanation is more
likely than any other. For a court to find that causation has been established on the balance of
probabilities when the scientific evidence is that causation cannot be established because of an
evidentiary gap, is to make a deliberately misinformed judgment based on intuition both as to
causation and as to overall responsibility for the loss. It is one thing for the court to be satisfied
on the balance of probabilities where there is some degree of doubt or uncertainty in the expert
evidence because the law requires only probability not certainty, but it is entirely different for a
court to resort to intuition about what probably happened where the expert evidence clearly
states that causation cannot be established.
2.1. The inferential approach to McGhee
As Lord Reid explained, the medical evidence in McGhee showed that brick dust can cause
dermatitis, and that the longer it is on the skin the greater the risk of developing dermatitis, but
could not explain why this is so. He conjectured that:
Plainly that must be because what happens while the man remains unwashed can have a
causative effect, although just how the cause operates is uncertain. I cannot accept the
view…that once the man left the brick kiln he left behind the causes which made him
178
liable to develop dermatitis. That seems to me quite inconsistent with a proper
interpretation of the medical evidence.39
According to Beever ‘the most natural reading’ of this passage is that ‘Lord Reid said that, on the
balance of probabilities, the defendant’s negligence caused the claimant’s injury’.40 He says that
this view was also supported by Lords Simon, Kilbrandon and Salmon so that they all ‘rightly or
wrongly, insisted that the claimant in McGhee could establish causation on the balance of
probabilities’.41 It is clear that there are two issues at stake here, first whether Lord Reid did find
causation to be established on the balance of probabilities rather than applying a new principle of
law and, second, whether the evidence in the case was actually capable of supporting such a
finding.
The first of these questions was resolved by the House of Lords in Fairchild where the majority
held that the facts of McGhee simply did not support a finding of causation on the balance of
probabilities and, given that Lord Reid had explicitly noted the distinction between the facts of
McGhee and Wardlaw, he cannot have intended to make a logically flawed finding of fact. This
means that the more interesting and more important question is the second question – was the
evidence in McGhee capable of supporting a finding of causation on the balance of probabilities?
Beever says that where there is scientific uncertainty ‘[c]rucially, the Court cannot simply side
with the defendant when the expert witnesses cannot or will not express their views on
probabilities’.42 There is some truth to this assertion but it needs to be heavily qualified. The
burden of proof in civil law rests on the claimant, so if the claimant is unable to persuade the
court of his claim on the balance of probabilities then it must fail. If the causal process is
understood i.e. it is known what constitutes a sufficient set, and the question is whether the
defendant’s negligence was necessary for the sufficiency of that set, then some pragmatism is
needed in law and care needs to be taken to see what the scientific evidence tells us and what are
McGhee (n10) 4-5 (Lord Reid), cited by Beever (n38) 466 (emphasis Beever’s).
Beever (n38) 466.
41 ibid.
42 ibid 471.
39
40
179
the limits of what it does and doesn’t say. But in cases involving an evidentiary gap problem the
scientific evidence is that it is impossible to say what constitutes a sufficient set of conditions.
This means that the experts’ refusal to say how likely it is that the negligence was necessary for
the sufficiency of the set is not a symptom of their adherence to a higher standard of proof than
the legal standard, but instead is based on the scientific impossibility of assigning a probability. In
other words, where it is theoretically possible to determine whether the negligence was a cause
but the expert witnesses are reluctant to express an opinion because they are observing a higher
standard of proof then it is still appropriate for the judge to assess his own degree of belief, but
where the evidentiary gap makes it impossible for the expert witnesses to express an opinion
there is no logical basis for the judge to have any degree of belief.
2.1.1 Scientific and legal standards of proof
Beever says that the medical experts in McGhee ‘said that it was not possible to prove…that the
defendant’s negligence caused the claimant’s injury, but, as the medical experts were not applying
the balance of probabilities, that does not imply that it was proved that the defendant did not
cause the plaintiff’s injury’.43 His argument therefore centres on the different standards of proof
in tort law and in science.
Courts have noted that standards of proof are higher in science than the civil law’s balance of
probabilities standard. Lord Prosser in Dingley v The Chief Constable, Strathclyde Police explained that:
Whether one uses the word ‘scientific’ or not, no hypothesis or proposition would be
seen as ‘proved’ or ‘established’ by anyone with any form of medical expertise merely
upon the basis that he had come to regard it as probably sound…And even if, in relation
to any possible proposition or hypothesis, such an expert even troubled to notice that he
had come to the point of regarding it as not merely possible but on balance ‘probable’,
then I think he would regard that point as one from which he must set off on further
43
ibid 467.
180
inquiry, and by no means as being (as it is in the courts) a point of arrival. Mere marginal
probability will not much interest him. But it must satisfy a court.44
Lord Phillips also noted the difference between legal and scientific standards in the later case of
Sienkiewicz v Greif (UK) Ltd:
When a scientific expert gives an opinion on causation, he is likely to do so in terms of
certainty or uncertainty, rather than probability. Either medical science will enable him to
postulate with confidence the chain of events that occurred, i.e. the biological cause, or it
will not. In the latter case he is unlikely to be of much assistance to the judge who seeks
to ascertain what occurred on a balance of probability.45
In the context of a negligence claim there is a clear place for pragmatism in the law. The courts
cannot postpone decisions until scientific certainty is achieved. As Lord Rodger said in Fairchild
‘the House must deal with these appeals on the basis of the evidence as to medical knowledge
today and leave the problems of the future to be resolved in the future’.46 Similarly, Miller has
highlighted that ‘civil courts cannot commission laboratory experiments or epidemiological
studies, but nor can they suspend a case until someone else does’.47 This echoes the argument
made in Chapter Two that whilst the legal test of causation ought to reflect a philosophically and
scientifically correct account of causation so far as is practicable, the law is also shaped by the
need to produce concrete decisions without delay.
But scientific or medical expert evidence cannot be written off as too demanding. Instead it is
essential that the courts engage with the evidence so that they understand what it does and does
not tell them in order to assess how persuasive the evidence is. This goes beyond identifying the
statistics to which the expert witnesses refer, and requires engagement on a qualitative level also.
As Laleng has suggested:
1998 SC 548, 603.
Sienkiewicz (n18) [9] (Lord Phillips).
46 Fairchild (n13) [124] (Lord Rodger).
47 Chris Miller, ‘Coal Dust, Causation and Common Sense’ (2000) 63 MLR 763, 769
44
45
181
an understanding of the similarities and differences between the two methods [legal and
scientific] is more likely to yield acceptable results than policy-based decisions that seem
to be based on little more than sympathy for victims and a possible generalised chemophobia.48
The previous chapter explained at length that the balance of probabilities standard of proof is a
‘belief probability’.49 The judge must be more convinced than not that the claimant’s account is
true. A belief probability involves a qualitative assessment of the evidence, and in order to form a
rational belief in causation in an individual case the court must assess not only the quantitative
evidence and ‘fact probability’, but also the sampling error probability, the reliability of the
studies from which the evidence is derived, and crucially must also attempt to personalise this to
the individual claimant’s case.
The previous chapter drew heavily on Barnes’ article which addresses the relationship between
legal and scientific standards of proof and seeks to dispel a number of misconceptions. He
explains:
Judges and lawyers first encountering statistical evidence want to believe that scientific
standards are tougher than legal standards. A court will reject an assumption that there is
no causal connection between an act and an injury if the evidence makes causation
“more likely than not.” A scientist will reject an assumption that there is no relationship
between two variables only if there is less than a five percent probability that the
statistical evidence showing a relationship is due to chance. The law appears willing to
accept no more than a forty-nine percent chance of error while science appears willing to
accept no more than a five percent chance of error.50
Per Laleng, ‘Sienkiewicz v Greif (UK) Ltd and Willmore v Knowsley Metropolitan Borough Council: A Material Contribution
to Uncertainty?’ (2011) 74 MLR 777, 793.
49 See Ch 3 section 3.2.
50 David W Barnes, ‘Too Many Probabilities: Statistical Evidence of Tort Causation’ (2001) 64 Law and Contemp.
Probs 191, 191.
48
182
He argues that ‘[t]his perception is incorrect, but hard to change’.51 As explained in chapter three,
the five percent chance of error in science concerns the sampling error probability (p-value), and
is not a belief probability so direct comparison is not possible. Furthermore, since a sampling
error probability reflects only the chance that the subjects of a study happened to be atypical it
can be calculated for a poorly-designed study as easily as for a well-designed study, so it would be
a doing a disservice to scientific experts to suggest that they are only concerned with the p-value.
2.1.1.1 The place of intuition in the balance of probabilities
In McGhee Lord Kilbrandon said that proof of causation ‘depends on drawing a distinction
between the possibility and the probability of the efficacy of the precautions. I do not find it easy
to say in the abstract where one shades into the other’.52 The line between possibility and
probability in the degree of belief is not a bright one because it necessarily involves a qualitative
assessment of the evidence which introduces an element of subjectivity. But it is the role of the
judge to decide on which side of the line a particular case falls. The qualitative aspect of the
standard of proof does not, however, mean that the balance of probabilities standard of proof
can be divorced from scientific evidence.
Stauch has criticised the balance of probabilities as being a crude test in comparison to statistical
evidence:
[W]hereas statistics derived systematically from our previous experience of similar cases,
provide us with a very accurate probability-weighting for each candidate [cause], the
balance of probabilities test attempts to perform the same operation by appealing crudely
to what we feel the likely cause to have been. The relevant feeling must, once again, derive
from our previous experience of similar cases, but this time in its rawest form.53
ibid.
McGhee (n10) 10 (Lord Kilbrandon).
53 Marc Stauch, ‘Loss of Chance in Medical Negligence’ (1997) 17 OJLS 205, 219.
51
52
183
There are a number of important observations to make in relation to this statement. The first is
that the balance of probabilities will necessarily seem less objectively measureable than statistical
probabilities because it is a qualitative concept. It is a belief probability so it measures the judge’s
degree of belief in a proposition so necessarily involves his assessment of how credible the
assertion is. Rather than being a weakness of the balance of probabilities standard of proof,
Barnes work has shown that the qualitative aspect is vital to assessing the credibility of the
scientific evidence and its value in an individual case. His work showed that in addition to the
fact probability and sampling error probability the court must weigh up other factors affecting
the reliability of the scientific study from which those probabilities were derived, and must
extrapolate to the individual case. While statistical probabilities may have an outward appearance
of objectivity, they only tell part of the story and they certainly do not provide a definitive answer
about causation in an individual case.
Additionally, Stauch presents us with a choice between statistical probabilities and the balance of
probabilities test and implies that by insisting on the balance of probabilities test the courts are
making a conscious and deliberate decision to rely on a feeling rather than on scientific and
statistical evidence. It is essential to remember that these two types of probability are not
mutually exclusive, indeed, as Barnes’ article showed, a rational belief probability must be based
on an assessment of the credibility of the scientific evidence. It is important for judges to
understand that the balance of probabilities standard requires them to form a belief based on the
available evidence, rather than ignoring the scientific evidence in favour of a ‘common sense’
judgment of causation.
2.1.1.2 Evidence of the evidentiary gap
As previously explained, it is essential to personalise the evidence as far as possible to the
individual claimant in order to form a rational belief about the causes of his injury and the
184
inferential approach loses sight of this vital step. The evidentiary gap means that it is not possible
to define a sufficient set, and this prevents meaningful personalisation of the evidence.
Adopting the inferential approach, Beever argues that ‘[c]rucially, the experts did not say that
there was no proof of a causal link between dust and dermatitis’.54 This is true, but it was not
suggested that the claimant was unable to prove the general link between brick dust and
dermatitis. The problem was proving the link between the negligent exposure to brick dust and the
dermatitis. Indeed, the fact that his dermatitis was caused by brick dust was what enabled the
court to say that the negligent exposure to brick dust had increased the risk of suffering
dermatitis. It is, however, a significant leap from saying that brick dust can cause dermatitis to
saying that the negligent dust was a cause of this claimant’s dermatitis.
As Mullany clearly explains, for a belief in the fact of causation to be rational it must be based on
scientific evidence since, ‘[t]he educative effect of the expert evidence makes appeals to common
sense notions of causation largely meaningless or produces findings which would not be made by
an ordinary person who had not been so instructed’. 55 In other words, if the ordinary person
looked at the events in McGhee without the assistance of expert evidence then they would
probably say that since brick dust is capable of causing dermatitis it makes sense to say that both
periods of exposure to brick dust caused this claimant’s dermatitis. But once they were shown
the medical evidence which said that dermatitis might be caused by an accumulation of
abrasions, or it might just start at a single abrasion, the ordinary person would not be able to say
that the negligent exposure contributed to the illness because they would see that it is as likely
that it was caused by a single abrasion during the innocent exposure.
The NESS analysis showed that the evidentiary gap surrounding the aetiology of the disease
means that it is not possible to define a ‘sufficient set’ of conditions. In cases such as McGhee and
Fairchild where there are a number of sources of the harmful agent and an evidentiary gap
preventing the definition of a ‘sufficient set’, it is simply not possible to form a rational belief
54
55
Beever (n38) 468.
Nicholas J Mullany, ‘Common Sense Causation – an Australian View’ (1992) 12 OJLS 431, 437.
185
that the negligence was necessary for the sufficiency of an actually occurring set of conditions
when it is not possible to say what a sufficient set of conditions consists of. No matter how far
the statistical and epidemiological evidence can be personalised to account for the individual’s
characteristics, and his working and living conditions, the evidentiary gap means that where there
is more than one source of the harmful agent it is not possible to form a rational belief that a
particular source was a cause. As the discussion of Barnes’ work showed, for a belief probability
to be rational it must be based on the available evidence and the evidence was clear that it was
impossible to know because a sufficient set could not be defined.
Although the majority in Fairchild rejected the inferential approach, Lord Hutton adopted it,
explaining that ‘I consider that this approach, whereby the layman applying broad common sense
draws an inference which the doctors as scientific witnesses are not prepared to draw, is one
which is permissible’.56 This is unjustifiable given that the problem was not that the scientific
evidence was inconclusive, but that it showed a clear evidentiary gap in being unable to define a
‘sufficient set’. This made it impossible to have a rational belief that the defendant’s negligence
was necessary for an actually occurring sufficient set. Weekes has argued that:
This reliance on “broad common sense” is perplexing. Expert evidence is tendered for
the purpose of proving or disproving matters which are beyond the ordinary knowledge,
sense and experience of the tribunal of fact. Thus when causation is beyond the powers
of even the experts, the inferential approach amounts to deliberately uninformed
guesswork.57
So it is one thing to say that ‘the law’s task is to determine the nature and form of evidence the
plaintiff should be permitted to adduce to prove the required historical link’,58 but it is not the
law’s task to find a link where the evidence is clear that the link cannot be established. As
Fairchild (n13) [100] (Lord Hutton).
Robert Weekes, ‘Not Seeing the Wood for the Trees – Risk Analysis as an Alternative to Factual Causation in
Fairchild’ (2003) 12 Nott LJ 18, 25.
58 Jane Stapleton, ‘Publication Review – Causation and risk in the law of torts: scientific evidence and medicinal
product liability’ (2000) 116 LQR 506, 506.
56
57
186
Hoffmann said in Fairchild, ‘however robust or pragmatic the tribunal may be, it cannot draw
inferences of fact in the teeth of the undisputed medical evidence’.59
2.2 The ‘doubling of the risk’ approach to proof on the balance of probabilities
One final aspect of the balance of probabilities standard of proof must be addressed and that is
the suggestion that the balance of probabilities can be satisfied by proof that the negligence more
than doubled the background risk so is the ‘most probable’ cause. This issue was raised in the
mesothelioma context in Sienkiewicz v Greif.60 The claim was brought by the daughter of a
mesothelioma victim who had been exposed to asbestos due to the negligence of the defendant
employer. Unlike the claimant in Fairchild she had not been exposed to asbestos by any other
employers, but she had been exposed to asbestos in the atmosphere of the area where she lived.
This ‘environmental exposure’ was not attributable to negligence; it was an ‘innocent’ source of
asbestos. The risk created by the environmental exposure was assessed at 24 cases per million,
and the risk created by the negligent exposure was assessed at 4.39 cases per million, so the
environmental exposure was by far the greater source of asbestos exposure. The defendant
advanced two arguments. First it was argued that the Fairchild test should not apply in this case
because, unlike Fairchild, there was only one negligent source of asbestos dust so it was suggested
that the case did not involve the same degree of uncertainty as Fairchild. The standard but-for test
should be applied, and, it was argued, causation could be proved on the balance of probabilities
only by proof that the negligent exposure had more than doubled the risk created by the
environmental exposure. Alternatively, the defendant argued that if the Fairchild test was
applicable in the circumstances, the increase in risk attributable to the defendant’s negligence
should only be regarded as ‘material’ if it at least doubled the background risk. The Supreme
Court held that the Fairchild test did apply because the evidentiary gap concerning the aetiology
of mesothelioma still exists and makes proof of causation impossible where there is more than
59
60
Fairchild (n13) [70] (Lord Hoffmann).
Sienkiewicz (n18).
187
one source of asbestos regardless of whether those sources are innocent or due to negligence.
The evidentiary gap was not avoided by the potential to measure the extent of the risk
attributable to each source.
Lord Phillips concluded in Sienkiewicz that statistical and epidemiological evidence was not
sufficient to establish causation on the balance of probabilities, explaining that it was considered
both inadequate and unreliable.61 The reliability of the epidemiological evidence is a factor that
affects the degree of belief in the conclusions that can be drawn from it. This issue was
addressed in chapter three, and although the statistical evidence in Gregg did not appear to be
reliable it is important to remember that where reliable statistical and epidemiological evidence is
available it can be a very valuable tool in proving causation. The issue that is specific to the
evidentiary gap is the inadequacy of the epidemiological evidence. Lord Phillips suggested that
‘so long as medical science is unable to demonstrate, as a matter of fact, the aetiology of
mesothelioma, data relating incidence to exposure is not a satisfactory basis for making findings
of causation’.62
It will be argued that Lord Phillips is right to say that epidemiological evidence cannot provide
adequate proof of causation where there is an evidentiary gap, but that his criticisms go too far.
It is important to observe that epidemiologists do not take evidence of a doubling of the risk as
proof of a causal connection so rejecting the doubling of the risk approach does not entail
rejection of epidemiological evidence. Indeed, once more is known about the aetiology of the
disease, epidemiological evidence may be very valuable in proving causation.
2.2.1 Doubling of the risk
While there is still an evidentiary gap in relation to mesothelioma it is arguable that
epidemiological evidence will continue to be inadequate proof of a causal link between the
individual claimant and defendant where there are multiple sources of asbestos exposure. As
61
62
ibid [97]-[98] (Lord Phillips).
ibid [97] (Lord Phillips).
188
already explained, the evidentiary gap pertains to the aetiology of the disease and specifically to
what constitutes a ‘sufficient set’. Because it is not possible to define a sufficient set of
conditions to cause mesothelioma it is not possible to prove that the negligence was a necessary
element of an actually occurring sufficient set of conditions. The balance of probabilities requires
the court to form a belief in the fact of causation, and unless it is possible to state what fact is
being proved, it is not possible to form a rational belief in that fact. As Steel and Ibbetson
explain:
The balance of probabilities rule is still a probabilistic rule in so far as the fact-finder’s
belief does not have to be a certain one. The main point, however, is that the
probabilistic aspect of the rule is a reflection of the fact-finder’s degree of belief, not the
content of that belief. This might be thought to be a purely philosophical point, but its
consequences are clear: if statistical evidence only generates belief in a probability, not a
fact, it cannot on its own satisfy the balance of probabilities rule.63
This can be illustrated by analysing Beever’s statistical approach to Fairchild. He argued that the
causal link between the claimant and a particular defendant/a particular source of asbestos could
have been proven on the balance of probabilities in Fairchild, but his argument is reliant on
statistical probabilities/relative frequency and, as seen above, this is not synonymous with proof
on the balance of probabilities which denotes belief probability i.e. is qualitative. Beever
illustrates his approach to the probability of causation in evidentiary gap cases through a
variation on the single-hit hunters case of Cook v Lewis.64 He introduces an element of evidentiary
uncertainty by proposing an example where there are three equally likely possibilities: the hunter
was hit by A, or was hit by B or was hit by both A and B. This is similar to the evidentiary gap
since the causal mechanism may be a single bullet or it may be ‘cumulative’ if it involved both
bullets. He says that the probability that A alone shot the hunter was 33.33 percent, the
probability that B alone shot the hunter was 33.33 percent, and the probability that A and B both
63
64
Steel and Ibbetson (n23) 464.
[1952] 1 DLR 1
189
shot the hunter was 33.33 percent. This means that the overall chance that A shot the hunter is
66.67 percent (achieved by adding together the probability that it was A alone and the
probability that it was A and B together). The same conclusion applies to B. Beever argues that
‘[t]he difficulty in Fairchild was caused, not by any inadequacy in the law’s traditional approach to
causation, but by misunderstanding the medical evidence and failing to understand this feature of
probability’.65
Beever engages in a forceful defence of this mathematical aspect of the calculation of
probabilities and his exposition of the mathematics is not challenged here. Unfortunately, as
discussed above in relation to Barnes’ paper on the different types of probabilities, this purely
mathematical probability is simply not what the balance of probabilities standard of proof
involves. The law is concerned to establish the fact that A shot the hunter, not the mathematical
probability that A shot the hunter. Unless the underlying causal process is known to involve
either a single bullet or an accumulation of bullets, it is not possible to form a rational belief that
A’s bullet was actually causally relevant. It is essential to understand what constitutes a ‘sufficient
set’ before being able to believe that A’s bullet was necessary for an actually occurring sufficient
set. This is part of the qualitative aspect of the balance of probabilities as a belief probability –
the mathematical probabilities are clear but it is not possible to extrapolate to the individual case
with any degree of credibility because of the evidentiary gap surrounding the aetiology.
Since Beever’s probability is not based on an understanding of the aetiology, it does not give rise
to a high degree of belief in the soundness of his statistical probability. Miller has explained
clearly that since the balance of probabilities relates to the degree of belief in a fact, where a
court is faced with statistical evidence it must form a degree of belief in the credibility of that
statistical evidence:
BOP [balance of probabilities] relates to the credibility or the degree of belief that the
fact-finder ascribes to the claimant’s overall attempt to attribute her harm to the
65
Beever (n38) 476.
190
defendant’s negligence. If there is uncertainty in the scientific evidence, that attribution
may have to be expressed in terms of risk, chance, odds, likelihood or other probabilistic
terminology. But the fact-finder must still gauge the overall credibility of that account –
thus, he may be 90% sure that the increased risk of the claimant’s injury is less than
30%.66
In the hunters scenario Beever provided, the mathematical probability that A was a cause was
66.67 percent, and the probability that B was a cause was also 66.67 percent. Despite this, if the
claimant brought a case against either one of them, this mathematical probability would not be
enough to create a belief that it is more probably than not true that he injured the claimant. This
is because the claimant must first prove that it was either a single bullet or two bullets that hit
him (what is a sufficient set – one bullet or two?), and then provide evidence that the defendant’s
bullet was necessary for the set that actually occurred.
2.2.2
Closing the evidentiary gap
The evidentiary gap may be overcome in the future as scientific knowledge developes so it is
important to note that it will not be a permanent problem. Once the evidentiary gap no longer
exists courts should revert to orthodox principles of damage and causation. McIvor criticises the
decision in Sienkiewicz, observing that many of the speeches suggest that ‘a claimant in a
mesothelioma case would always need to be able produce evidence which identifies the precise
point in time at which the genetic process triggering the malignant mutation was initiated’.67 This,
she says, reflects a standard of proof much closer to the criminal law requirement of proof
‘beyond all reasonable doubt’ than the civil law balance of probabilities standard. It is not
necessary to prove the precise point in time when the mesothelioma was triggered, but it is
necessary to prove that the particular causal process was more likely than not the process that
Chris Miller, ‘Liability for negligently increased risk: the repercussions of Barker v Corus UK (Plc)’ (2009)8 LP&R
39, 51.
67 Claire McIvor, ‘Debunking some judicial myths about epidemiology and its relevance to personal injury litigation’
(unpublished conference paper, Obligations VI: Challenging Orthodoxy conference, Canada 18 th-20th July 2012) 12.
66
191
occurred. This is currently not possible; as Lord Phillips explained, while the single fibre theory
has been discredited it is still not known whether the fibres have a synergistic interaction, nor at
what stages in the development of the disease the fibres play a part. It is not possible to show
that one process is more likely than the others. But if/when science enables us to say that causal
potency does depend on the quantity/type of asbestos, or on intensity of exposure, or that
earlier/later exposures are more likely to cause mesothelioma, then it may be possible to prove
causation on the balance of probabilities.
McIvor explains that epidemiology is more far-reaching than Lord Phillips suggests it to be. In
Sienkiewicz he stated that:
Epidemiology is the study of the occurrence and distribution of events (such as disease)
over human populations. It seeks to determine whether statistical associations between
these events and supposed determinants can be demonstrated. Whether those
associations if proved demonstrate an underlying biological causal relationship is a
further and different question from the question of statistical association on which the
epidemiology is initially engaged.68
But as McIvor argues, ‘[w]hile statistics are certainly an important tool used by epidemiologists,
they are no more than a tool and epidemiologists are more than just statisticians’.69 She explains
that epidemiologists are trained not only in statistical techniques but also in research
methodologies and medicine, so not only do they design and carry out studies, they then ‘use
various sophisticated techniques for interpreting the data precisely in order to determine whether
any indicated statistical associations provide evidence of an underlying biological relationship of
cause and effect’.70 As she suggests, if epidemiologists were used as expert witnesses they would
be able to draw meaningful and often useful conclusions from epidemiological evidence. Miller
has also criticised the courts for being excessively dismissive of epidemiological evidence.
Sienkiewicz (n18) [80] (Lord Phillips).
McIvor (n67) 15.
70 ibid.
68
69
192
Because the evidence is expressed in statistical or probabilistic terms, he says that courts have
been unwilling to draw causal inferences but, he argues, the Bradford Hill criteria used by
epidemiologists to assess the reliability of causal inferences ‘are unashamedly pragmatic and if
judges were to reject epidemiological evidence which was deemed to have satisfied those criteria,
then such opposition would be hard to defend’.71
This section has shown that where there is an evidentiary gap surrounding the aetiology of a
disease which prevents the determination of a ‘sufficient set’ of conditions, and there are
multiple sources of the harmful agent, it is not possible for a claimant to prove on the balance of
probabilities that a particular source of the relevant harmful agent was a cause. Since the content
of a sufficient set is unknown, even in theory, it is not possible to show that the defendant’s
negligence was necessary for the sufficiency of any set of conditions that actually occurred. The
claimant is required to prove the fact of causation, and the fact-finder must believe this to be
more probably than not true in the sense that he must believe in the credibility of the evidence.
The balance of probabilities standard of proof relates to belief probability which involves a
qualitative assessment of the fact probability, of the sampling error probability, taking into account the
design and conduct of the scientific study in question and of how far this can be extrapolated to
the individual case of the claimant. The balance of probabilities has to be a belief probability
because the fact-finder has to ask the question of the extent to which a study based on a given
population can inform us about an individual with the claimant’s characteristics in the claimant’s
circumstances.
This means that the fact-finder’s belief would be irrational if he chose to disregard clear scientific
evidence as to causation and replace it with a ‘common sense’ view drawing on uninformed
intuition. It also means that one cannot translate statistical probabilities directly on to the balance
of probabilities standard of proof, so proof of a ‘doubling of the risk’ or a purely mathematical
71
Chris Miller, ‘Causation in personal injury: legal or epidemiological common sense?’ (2006) 26 LS 544, 566.
193
probability calculation such as Beever advocates, can only give rise to a probability of causation
and the fact-finder must still ask what that actually tells him about the individual case. Without a
scientific understanding of a sufficient set, one can only hold a low belief in the capacity of such
statistical probabilities to provide meaningful information about an individual case. This means
that the leap made by the majority in McGhee from finding that the defendant’s negligence had
‘materially increased the risk of harm’ to concluding that his negligence had made a material
contribution to the harm itself, cannot be accepted as drawing an inference about factual
causation. As Nolan has argued, ‘[t]his conflation of material increase in risk and material
contribution was a fairly blatant judicial sleight of hand’.72 Indeed, this interpretation was
disapproved in Fairchild where it was held that the evidentiary gap prevented proof of causation,
but that to assist claimants facing this insuperable problem of proof the defendant would be held
liable on the basis that he had caused a material increase in the risk of harm. Under Fairchild, as
developed in Barker, proof of a material increase in the risk of harm does not prove that the
negligence was a cause of the harm. Instead the basis of liability has changed from liability for
causing the claimant’s loss, to liability for increasing the risk of the claimant’s loss. Liability is no
longer based on the fiction of a causal link, but explicitly based on the fact that the defendant
increased the risk and this is enough to engage liability. The following section will therefore turn
to this solution to the evidentiary gap and its place within a corrective justice-based account of
negligence.
3. Risk of harm as the gist of the negligence action: Barker v Corus (UK) plc73
In order for negligence liability to be consistent with corrective justice, there must be a causal
link between the defendant’s negligence and the claimant’s loss. The previous section has shown
that the evidentiary gap prevents proof of a causal link between the negligence and the physical
Donal Nolan, ‘Causation and the Goals of Tort Law’ in Andrew Robertson and Tang Hang Wu (eds.) The Goals of
Private Law (Hart Publishing 2009) 168.
73 Barker v Corus (UK) plc [2006] UKHL 20, [2006] 2 AC 572.
72
194
harm suffered by the claimant. The majority of the Supreme Court decision in BAI v Durham
held that the Fairchild rule is a special rule whereby a defendant whose negligence materially
increased the claimant’s risk of mesothelioma is deemed to have caused the mesothelioma. 74
However, an alternative approach is to reconceptualise the actionable damage in these cases as
exposure to the risk of harm rather than the physical harm itself since the claimant is able to
prove that the defendant’s negligence caused this ‘loss’. This was the approach Lord Hoffmann
purported to follow in Barker v Corus, and in his dissent in BAI v Durham Lord Phillips
maintained that this was the correct interpretation of Barker. It will be argued here, however, that
this this approach is still not consistent with corrective justice. There are two strands to this
argument. First, Lord Hoffmann insisted that the risk must have materialised into physical harm
before the claimant would be able to recover for the exposure to the risk of harm, and this
requirement means that it is still the physical harm that forms the gist of the action. Secondly,
even if the physical harm requirement were to be removed, risk exposure lacks the independent
moral significance to form damage in the negligence inquiry.
3.1 The decision in Barker v Corus
The claim in Barker v Corus raised the question of whether the Fairchild principle was applicable
when the claimant had been exposed to asbestos not only due to the negligence of his former
employers including the defendant, but also by himself during a period of self-employment. In
Fairchild the House of Lords had said that any background exposure to asbestos was so small
that it could be effectively discounted so all of the claimant’s exposure was attributable to former
employers.75 This meant that, although the evidentiary gap prevented the claimant from proving
that a particular defendant was a cause of his mesothelioma, it was accepted that his loss was
certainly attributable to somebody’s negligence. The claimant in Barker was not an ‘innocent
BAI v Durham (“Trigger Litigation”) (n4) (Lords Mance, Kerr, Clarke and Dyson).
The question of whether it was correct to ignore the background risk, and the related ‘innocent claimant’
argument, will be addressed in section 4.1.
74
75
195
claimant’ in this sense because the period of self-exposure meant that he could not say that his
loss had definitely been caused by somebody else’s negligence. Given the possibility that the
claimant had caused his own loss the fairness of shifting the burden of the loss onto a negligent
employer was less apparent. As an alternative argument, the defendant claimed that if the
Fairchild principle was applicable in this case then liability ought to be several rather than joint
and several.
The House of Lords held that the Fairchild principle did apply even though the claimant himself
was responsible for some of the asbestos exposure. The court did, however, accept the
defendant’s second line of argument and held that where the Fairchild principle is applied liability
ought to be several rather than joint and several. Lord Hoffman, with whom Lord Scott, Lord
Walker and Baroness Hale agreed on the applicability of the Fairchild principle, based his finding
on causal principles:
For this purpose it should be irrelevant whether the other exposure was tortious or nontortious, by natural causes or human agency or by the claimant himself. These
distinctions may be relevant to whether and to whom responsibility can also be
attributed, but from the point of view of satisfying the requirement of a sufficient causal
link between the defendant’s conduct and the claimant’s injury, they should not matter.76
This echoes the argument made in Chapter Two that causation is a factual question of
involvement so it is irrelevant to the causation inquiry whether a candidate cause is innocent or
negligent. Instead this characterisation is relevant to the attribution of responsibility, of which
causal involvement is just one element. In particular, the duty of care and principles of legal
causation define the limits of moral responsibility.
Lord Rodger’s dissent in Barker focused instead on balancing the interests of the claimant and
defendant to achieve a just solution. He considered that the Fairchild principle had been
motivated by the demands of fairness and the need to avoid injustice to the claimant. Assuming
76
Barker (n73) [17] (Lord Hoffmann).
196
that liability remained joint and several, he was concerned that applying the Fairchild principle
where the claimant was responsible for a period of asbestos exposure would tip the balance too
far in his direction, resulting in injustice to the defendant.77 However, given that the majority
favoured several liability he accepted that ‘the balance of potential injustices favour applying the
Fairchild exception’.78
Section Four will look at broader policy arguments and considerations of fairness. The aspect of
the decision that is considered here is the reasoning behind the decision that liability ought to be
several. The House of Lords were in agreement that the extent of the defendant’s contribution
to the total risk was an appropriate measure for the apportionment of liability that would take
place. It remained a requirement that the claimant must have suffered physical harm in the form
of mesothelioma, meaning that the risk must have materialised. The value of the total risk was
therefore said to be the same as the value of the physical harm. The defendant would
compensate the claimant for a portion of the value of the physical harm, with the extent of that
portion measured by his contribution to the total risk. Lord Scott explained that this
apportionment would be based on the duration of the exposure for which each defendant was
responsible compared with the total duration of exposure to asbestos. It might also take into
account the intensity of the exposure and, since different types of asbestos increase the risk more
than others, the type of asbestos dust.79
The majority were divided in the reasons they gave for this decision. One approach was to retain
the physical harm, mesothelioma, as the actionable damage or ‘gist’ of the negligence action, but
to apportion liability according to contribution to risk on the basis that this achieved a fair
balance between the competing interests of the claimant and defendant. The second approach
was to hold that where the Fairchild principle is applied the gist of the action is no longer the
ibid [98]-[101] (Lord Rodger).
ibid [101] (Lord Rodger).
79 ibid [62] (Lord Scott).
77
78
197
physical harm but the exposure to risk itself. Baroness Hale favoured the first justification, that
apportionment was a ‘fair’ solution:
The law of tort is not (generally) there to punish people for their behaviour. It is there to
make them pay for the damage they have done. These Fairchild defendants may not have
caused any harm at all. They are being made liable because it is thought fair that they
should make at least some contribution to redressing the harm that may have flowed
from their wrongdoing. It seems to me most fair that the contribution they should make
is in proportion to the contribution they have made to the risk of that harm occurring.80
The relevance of risk, in her approach, was no more than that it provided a ‘sensible basis’ on
which liability could be apportioned.81
In contrast, risk took on much greater significance in Lord Hoffmann’s justification for applying
several liability. Under the Fairchild principle a defendant is held liable on the basis that his
negligence caused a material increase in the risk of harm. This meant, in his opinion, that the
increase in risk, rather than the physical harm, ought to form the gist of the negligence action:
Consistency of approach would suggest that if the basis of liability is the wrongful
creation of a risk or chance of causing the disease, the damage which the defendant
should be regarded as having caused is the creation of such a risk or chance.
He continued to explain:
If that is the right way to characterize the damage, then it does not matter that the disease
as such would be indivisible damage. Chances are infinitely divisible and different people
can be separately responsible to a greater or lesser degree for the chances of an event
happening.82
This solution has an appearance of consistency with corrective justice since the defendant is
ostensibly being held liable only for the loss that he has been proved to have caused i.e. exposure
ibid [127] (Baroness Hale).
ibid [126] (Baroness Hale).
82 ibid [35] (Lord Hoffmann).
80
81
198
to the risk of mesothelioma. If the actionable damage is the risk of mesothelioma then the
centrality of the causation requirement is maintained since the claimant must still prove on the
balance of probabilities that the defendant’s breach of duty caused this damage. On closer
inspection, however, this approach is not consistent with corrective justice. There are two
reasons for this. First, the continued insistence that the claimant must have developed
mesothelioma undermines the claim that the gist has been changed to the risk of mesothelioma.
In other words, the damage for which the claimant was compensated was still the mesothelioma
itself, but liability was apportioned to reflect the degree of uncertainty surrounding proof of the
causal link. Secondly, even if the courts were to abandon the requirement that the claimant must
suffer the physical harm, exposure to risk cannot be considered to be damage within corrective
justice because it adds nothing to the breach inquiry. Conduct is characterised as wrongful if it
exposes the claimant to an unreasonable risk of harm. Corrective justice does not punish this
wrongful behaviour, but requires the wrongdoer to repair the loss when his wrongdoing causes
harm to another. If a defendant commits a wrong by exposing a claimant to an unreasonable risk
of harm but this risk does not materialise then he is not liable because there is no loss for him to
correct. If the law was to hold that exposure to risk does constitute harm then the requirement
of damage would effectively be lost and the basis for liability would not be corrective justice but
a retributive form of justice.
3.2 Analysis of the ‘risk as gist’ approach
As noted above, Lord Hoffmann explained that ‘[c]onsistency of approach would suggest that if
the basis of liability is the wrongful creation of a risk or chance of causing the disease, the
damage which the defendant should be regarded as having caused is the creation of such a risk
of chance’.83 Even Lord Rodger, who ultimately did not support the risk-based analysis, thought
that the ‘dominant aim’ pursued by changing the gist of the action to the risk of harm was ‘to
83
ibid [35] (Lord Hoffmann).
199
secure internal consistency between the basis of liability and the nature of the damage’. 84
However, writing after the House of Lords’ decision in Fairchild, Stapleton has said:
The majority of the House…firmly rejected the use of legal fictions. Refreshingly, they
did not adopt any of a range of possible pretences. One would have been that the
McGhee/Fairchild rule related to a special duty, such as a duty to avoid exposing the victim
to an increased risk of mesothelioma which would have constituted a special type of
wrong twinned with a special sort of ‘damage’.85
It will be argued that, as Stapleton suggests, the ‘risk as damage’ approach rests on a fiction and
conceals the fact that the gist is actually still the physical harm and that the apportionment
exercise reflects the remaining doubt over causation.
It will also be argued that Lord Hoffmann’s approach introduces confusion to the interaction
between the damage, causation, and quantification elements of the negligence inquiry. Stapleton
has said that it ‘cannot be over-emphasised that the formulation of the “damage” forming the
gist of the action defines the causation question. Logically one can only deal with causation after
one knows what the damage forming the gist of the action is’.86 This is problematic in Lord
Hoffmann’s approach because he suggests that the gist of the action is ‘risk’ which is a divisible
damage, but the claim is only actionable when the claimant has suffered the physical outcome
harm, mesothelioma, which is indivisible. Where damage is divisible there should be
‘apportionment’ of liability as a matter of causal responsibility since each source of the harmful
agent causes a portion of the total loss. Where damage is indivisible ‘apportionment’ (vis-à-vis
the claimant; under joint and several liability the defendant can seek contribution from his codefendants) takes on a different meaning. This exercise would take place at the liability stage as
an exception to the usual rule of joint and several liability. The location of this ‘apportionment’
ibid [84] (Lord Rodger). His aim was to maintain ‘a consistency of approach with the main body of law on
personal injuries’ (ibid [84]).
85 Stapleton, ‘Lords a’leaping’ (n24) 290.
86 Jane Stapleton, ‘The Gist of Negligence: Part 2 The Relationship Between “Damage” and Causation’ (1988) 104
LQR 389, 393.
84
200
exercise is therefore confused in Lord Hoffmann’s approach which notionally treats a divisible
harm as the gist while insisting on the occurrence of an indivisible harm and using this as a
measure for the divisible harm.
3.2.1 The inconsistency of the physical harm requirement
Although Lord Hoffmann said that the gist of the action was the risk of mesothelioma, it was
still a condition of liability that the claimant must have developed mesothelioma. This
requirement of physical harm may be a practical way of limiting the number of cases where the
Barker principles applies, but it is conceptually problematic because it means that the gist was not
actually redefined as the risk of harm. Lord Hoffmann stated:
Although the Fairchild exception treats the risk of contracting mesothelioma as the
damage, it applies only when the disease has actually been contracted. [Counsel for the
defendant] was reluctant to characterise the claim as being for causing a risk of the
disease because he did not want to suggest that someone could sue for being exposed to
a risk which had not materialised. But in cases which fall within the Fairchild exception,
that possibility is precluded by the terms of the exception. It applies only when the
claimant has contracted the disease against which he should have been protected.87
This reasoning is inadequate to support his assertion that the gist could be redefined as the risk
of harm whilst still requiring the claimant to have developed mesothelioma. If the gist of the
action is exposure to risk but the risk must have materialised before the action can be brought
then there is an internal inconsistency between the content of the rule and the scope of
application of the rule. Far from justifying the risk as damage approach, this inconsistency
undermines the rule. Indeed, Beever says that ‘the refusal to compensate the defendant’s former
employees unless they suffer mesothelioma reveals that the actionable damage is the
mesothelioma and the consequences thereof, and not the risk of mesothelioma…The idea that
Barker (n73) [48] (Lord Hoffmann). Affirmed in Rothwell v Chemical and Insulating Co Ltd [2007] UKHL 39, [2008] 1
AC 281.
87
201
these cases involve liability for risk creation is an illusion’.88 The requirement that the claimant
must have developed mesothelioma undermines the ‘risk as damage’ approach so severely that it
cannot simply be explained away by Lord Hoffmann as a requirement for the application of the
Fairchild principle. If he was willing to adopt the ‘risk as damage’ approach because he prioritised
‘consistency of approach’ then the physical harm requirement should have been abandoned in
order to actually achieve consistency of approach. As noted by Scherpe, ‘[n]othing is gained by
replacing one fiction with another but much can be lost’.89
3.2.2 Physical harm and risk: a shift in perspectives
There is a danger that this argument seems to focus solely on the appearance of inconsistency that
arises when the court says that the gist is the risk of mesothelioma whilst still insisting that the
claimant must have developed mesothelioma. Examining the nature of ‘risk’, however, will show
that this problem is not just apparent, but goes to the essence of the concept of risk. The
following section will look more closely at what ‘risk’ is, to show that risk is a forward-looking
concept and that, properly understood, the extent of the risk created by a particular defendant is
measured at the time it is created. There is an inherent mismatch between risk as a forwardlooking concept and the backward-looking role in which risk is being cast in Lord Hoffmann’s
approach. Here the extent of the defendant’s contribution to risk is only measurable once the
risk has materialised and all the other sources of the same kind of risk have been identified. This
means that the extent of the risk that the defendant created will vary in size depending on how
many other sources of asbestos the claimant was exposed to, whereas risk is actually a forwardlooking concept and measureable independently of other sources of risk. This mistaken use of
risk seems attributable to the subtle shift in the phraseology of the Fairchild exception which
Beever (n38) 487. See also Nolan (n72) 178; Lara Khoury ‘Causation and risk in the highest courts of Canada,
England and France’ (2008) 124 LQR 103, 126.
89 Jens Scherpe, ‘A New Gist?’ (2006) 65 CLJ 487, 488.
88
202
originated as ‘material increase in risk’ in McGhee and has morphed into ‘material contribution to
risk’ in Barker.
3.2.2.1 What is ‘risk’?
Perry has said that ‘[i]n ordinary language conduct is typically said to be risky when it gives rise to
a chance of a bad outcome of some kind. The concept thus involves two main elements: first, a
notion of chance or probability, and second, a notion of harm’. 90 It involves a state of
uncertainty as to the future. The measure of a particular risk is a product of the extent of the
possible harm that may be produced, and the probability of this harm materialising. Notably,
probability is a factor in calculating risk but it is not synonymous with risk. In Risks and Legal
Theory, Steele set out a range of different meanings assigned to the word ‘risk’. She explained that
for some theorists we can only truly speak of ‘risk’ when the probability of the relevant outcome
can be calculated. If the probability cannot be calculated then the situation ought to be described
as involving uncertainty, indeterminacy, or ignorance. Others use the word ‘risk’ to describe any
situation involving a hazard regardless of whether the probability of harm can be calculated. She
says that the latter approach ‘reflects an increasing colloquial and theoretical understanding of
risks as threats, rather than as statistical probabilities’.91 Notably the notion of risk relates to the
future in all of these forms. ‘Risk’ is also sometimes used as shorthand for a particular decisionmaking method. In this sense ‘risk’ does not refer to a danger that necessitates a decision-making
method in order to be managed, but refers to a way of approaching a problem or danger. 92 Steele
also notes that ‘risk’ and ‘probability’, ‘though related, are distinct terms’.93 Risk will sometimes
depend on a formal assessment of probability, and always depends on the likelihood of an
outcome. Probability can be used for analysis of past events, e.g. what is the probability that X
Stephen R Perry, ‘Risk, Harm, and Responsibility’ in David G Owen (ed.) Philosophical Foundations of Tort Law
(Oxford University Press 1995) 322.
91 Jenny Steele, Risks and Legal Theory (Hart Publishing 2004) 7.
92 ibid 7.
93 ibid 19.
90
203
caused Y?, so probability does not always entail a risk assessment. In contrast, risk generally
concerns the future although it can be used to debate the past,94 for example to assess what
would in the past have been a rational way to behave.95 Once again we see that probability is one
element of risk, but it is not synonymous with risk. This means that it is crucial that the two
terms, ‘risk’ and ‘probability’, are not used interchangeably.
So while probability may be used to analyse past events, risk is a forward-looking concept. It
describes a situation of uncertainty as to whether a particular outcome will occur. When a risk
materialises it causes an outcome so it is no longer a ‘risk’ but a cause. Of course there may be
uncertainty as to what caused a particular event i.e. uncertainty as to which risk(s) materialised,
and as Steele noted above, probability can aid the analysis of causation. It would be incorrect to
say that if a risk has materialised it has contributed to the risk of the outcome – this is just
circular. In other words a ‘risk’ is only a ‘risk’ while the outcome is still prospective. Once the
outcome occurs then the relevant question is whether the risk materialised and made a causal
contribution. Weekes explains:
By definition the concept of risk is a limitless one. The meaning of a factual cause is
binary: a given activity either causes or does not cause the damage in question. The limit
of a factual cause is where a given activity does not result, in whole or in part, in the
certain damage. Risk, however, is defined by degree. A risk which is proven to have
resulted in damage is of course a risk that has been realized, or more properly, “a
cause”.96
Beever explains the same idea through the analogy of a raffle that is held with 100 tickets each
costing £2 and a single prize of £1,000 (where the ticket has no value other than the fact it
provides a chance to win the £1,000 prize). Before the raffle is drawn, he explains, ‘your ticket is
ibid 20.
ibid 9.
96 Weekes (n57) 27.
94
95
204
worth £10, that being onehundredth of £1,000. But imagine now that the raffle is held and you
do not win. How much is your ticket worth now? It is worthless’.97
3.2.2.2 Can risk constitute damage?
Lord Hoffmann held in Barker that mesothelioma is no longer the actionable damage for which
the claimant recovers. Instead the actionable damage for which the claimant recovers is the risk
of mesothelioma to which the defendant exposed him. However, the claimant cannot claim in
respect of this exposure to the risk of mesothelioma until he actually develops mesothelioma,
and on Lord Hoffmann’s reckoning it is then possible to establish how much the defendant
contributed to the total risk to which the claimant was exposed. It will be argued that what this
approach actually achieves is to make the defendant liable for the mesothelioma itself, but to
discount the extent of his liability to reflect the uncertainty over whether the risk he created was
the risk that actually materialised. The method used to calculate the appropriate discount is the
probability that it was the defendant’s risk rather than another source of risk that materialised. It
cannot be the case that exposure to risk is the damage that is being compensated here because
risk is forwards-looking and only exists before the outcome occurs. After the outcome has
occurred, the risk that the defendant created either is or is not a cause of that outcome. Since risk
is forwards-looking the extent of the risk is measurable at the time of exposure to the risk and its
value is fixed at that time. In contrast, in Lord Hoffmann’s approach, the measure of the
defendant’s liability will vary with each subsequent exposure of the claimant to the same type of
risk. This variability shows that it simply cannot be ‘risk’ that is actually being compensated in
Lord Hoffmann’s approach. Instead he is awarding partial compensation for the mesothelioma
to reflect the probability that the defendant caused it. But we must remember that ‘probability’ is
not synonymous with ‘risk’ – probability is one element of risk calculation, but whereas
97
Beever (n38) 486
205
probability can look both forwards and backwards, risk only looks forwards. This is the
argument that will be elaborated in more detail here.
3.2.2.2.1
Calculating risk and probability
As explained, risk is a forward-looking concept and is a product of the extent of possible harm
and the probability of that harm occurring. When an employer negligently exposes an employee
to asbestos dust he exposes the employee to the risk of a number of asbestos-related diseases
such as lung cancer, asbestosis and mesothelioma. The risk can be measured at that time based
on the extent of the loss the employee would suffer if the risk were to materialise and the
probability of it materialising. If the employee chose to take out a health insurance policy it
would take into account the increased risk of the employee suffering these illnesses in the future
– it is the increase in the probability of the disease that we tend to focus on because the extent of
the loss does not vary. This focus on the probability aspect of risk may explain why Lord
Hoffmann uses the term ‘risk’ when he is actually referring to ‘probability’.
The relevant risk in these cases is the risk of mesothelioma. To provide a concrete example, in
Sienkiewicz the risk of mesothelioma created by the environmental exposure to asbestos was 24
cases per million. The risk of mesothelioma created by the occupational exposure to asbestos
was 4.39 cases per million.98 The occupational exposure was therefore said to have increased the
risk of the claimant developing mesothelioma by 18 percent.99 ‘Risk’ here is a predictive tool that
informs us how likely it is that the employee will develop mesothelioma. Based on the duration
and extent of the exposure to asbestos dust, and the type of dust, it tells us how much more
likely it is that the employee will develop mesothelioma than if she had not been exposed to each
source of asbestos dust. If, hypothetically, the claimant had then been exposed to the same kind
of risk by a later employer this would not alter the fact that the defendant had increased the risk
For the purposes of this example it is assumed that these figures are accurate although McIvor notes that the
judge calculated them himself from limited evidence and without the help of epidemiologist expert witnesses (n67).
99 (4.39 / 24) * 100 = 18.29
98
206
of mesothelioma by 18 percent. The risk that was created by the defendant does not increase or
diminish when the employee is exposed to the same kind of risk by subsequent employers.
This can be explained through an equivalent variation on Beever’s raffle analogy. In that example
100 tickets were sold for a raffle with a single prize of £1,000. If A has one ticket, his ticket is
worth £10 since that is one hundredth of £1,000. In terms of ‘risk’, the value of the ‘risk’ is £10
because it is a product of the value of the possible outcome, £1,000, and the probability of that
outcome, 1/100. If B gives his ticket to A, he has added another 1/100 to the probability that A
will win £1,000. If C has two tickets and gives these to A, he has added another 2/100 to the
probability that A will win £1,000. A’s initial ticket is still worth £10 because it has a 1/100
chance of winning, the ticket he got from B is still worth £10 because it has a 1/100 chance of
winning.
Lord Hoffmann said that ‘the basis of liability is the wrongful creation of a risk or chance of
causing the disease’.100 If this was the case then the value of the risk created by a particular
defendant would be fixed at the time of the exposure to that risk and would not vary. But this is
not how Lord Hoffmann’s approach works, so liability cannot actually be being imposed in
respect of risk. Starting with the raffle analogy, the equivalent to developing mesothelioma would
be A winning the raffle prize of £1,000. The 99 losing tickets are worthless; the 1 winning ticket
is worth £1,000. In the first scenario where A only has one ticket, this is the ticket that has been
drawn and Lord Hoffmann would say that A contributed the whole of the ‘risk’ of winning. In
the second scenario where A had another ticket given to him by B, Lord Hoffmann would say
that A contributed 50 percent of the ‘risk’ of winning, and B contributed the remaining 50
percent. And in the final scenario where A was given a further two tickets by C, Lord Hoffmann
would say that A contributed 25 percent of the ‘risk’ of winning, B contributed 25 percent of the
‘risk’ of winning, and C contributed 50 percent of the ‘risk’ of winning. So the extent of the ‘risk’
created by A varies, as does the extent of the ‘risk’ created by B. But we know that risk is a
100
Barker (n73) [35] (Lord Hoffmann).
207
forwards-looking concept and that its measure does not depend on other subsequent exposures
to the same kind of risk, so Lord Hoffmann cannot actually have made ‘risk’ the actionable
damage. Once again the problem appears to be that he treats ‘risk’ as synonymous with
‘probability’ perhaps because of the role that probability plays in allowing us to calculate risk. As
noted above, we also have a tendency to focus only on the probability element of risk when
comparing risks of the same outcome because the value of the possible outcome is a constant
and the risk-creator increases the probability of this outcome occurring. Since the outcome has
occurred, A has won £1,000, one of his tickets is worth £1,000 and the remaining three are
worth nothing. There is a 1/4 probability that it is his own ticket, a 1/4 probability that it is the
ticket he received from B and a 2/4 probability that it is a ticket he received from C. If we made
him divide up the prize according to these probabilities, keeping £250 for himself, giving £250
to B and £500 to C we would be allocating the prize according to the purely statistical probability
that each person was responsible for the winning ticket. It may be that in cases of evidentiary gap
this is the best the court can achieve, to apportion liability according to the statistical probability
that the defendant caused the disease. But since causation of the physical outcome cannot be
proven on the balance of probabilities, the decision to impose proportionate liability for possible
causation cannot be based on corrective justice. If we want to recognise exposure to the risk of
mesothelioma as actionable damage in itself, then we need to understand that this damage occurs
at the time of the exposure and has value that is independent of other exposures to the same
kind of risk.
3.2.2.2.2
‘Increase in risk’ or ‘contribution to risk’?
This conflation of risk and probability is hidden in the subtle, yet significant, shift in the wording
of the Fairchild principle. The Fairchild principle is expressed both as a test of ‘material increase in
the risk’ of harm, and of ‘material contribution to the risk’ of harm. These phrases seem similar
208
but they actually mean very different things. The Court of Appeal in Williams v University of
Birmingham asked whether:
…the law requires a judge to do a comparison between the two (or more) sources to see
if the wrongful exposure at the hands of the defendant is, by comparison with the other
exposures, too insignificant to be taken into account. Or, is it enough for a judge to find
as a fact that the wrongful exposure at the hands of the defendant materially increased
the risk that the victim would contract mesothelioma?101
Asking whether the defendant’s negligence materially increased the risk of harm reflects the
forwards-looking concept of risk. An increase in the risk is measurable at the time of the
exposure and if it is a ‘material’ increase then this will not change because the measure of the risk
is unaffected by later exposures. In contrast, asking whether the defendant’s negligence made a
‘material contribution to the risk’ does require a comparison with the other sources of risk. The
contribution that the defendant made to the total ‘risk’ will be greater or smaller depending on
how many other employers exposed the claimant to asbestos, over what period and with what
intensity. As discussed above, this variability in the measure of the so-called ‘risk’ shows that
actually we have shifted to measuring probability of causation. This change in the terminology
seems to have occurred in academic discussion following Fairchild.102 It was firmly entrenched by
the time of Sienkiewicz where Lord Phillips said ‘the only circumstances in which a Court will be
able to conclude that wrongful exposure of a mesothelioma victim to asbestos did not materially
increase the victim’s risk of contracting the disease will be where that exposure was insignificant
compared to the exposure from other sources’.103 Whether conscious or unconscious, it is
possible that it reflects a desire to draw a parallel with the Wardlaw test of ‘material contribution
to harm’ that is a standard test for causation. The discussion above shows that courts must be
[2011] EWCA Civ 1242, [2012] PIQR P4 [69] (Aikens LJ).
See for example Stapleton ‘Lords a’leaping’ (n15).
103 Sienkiewicz (n18) [11] (Lord Phillips). Although the Court of Appeal in Williams concluded that it does not require
a comparative exercise (n101) [72] (Aikens LJ).
101
102
209
careful when making subtle changes to the wording of a legal test because this can open the way
for the test to take on a very different meaning.
3.2.3 Confusion in the ‘apportionment’ issue
Beever has highlighted another fallacy in Lord Hoffmann’s approach, explaining that if the
‘harm’ is divisible then responsibility for it is already ‘apportioned’ as a matter of causation
without the need subsequently to apply several liability.104 Lord Hoffmann’s approach is
therefore problematic because he suggests that the gist of the action is ‘risk’ which is a divisible
damage, but the claim is only actionable when the claimant has suffered the physical outcome
harm, mesothelioma, which is indivisible. This means that he collapses the distinction between
defining the damage that the defendant caused and determining the extent of the defendant’s
liability for that damage.
As argued in Chapter One, it is important to maintain a clear distinction between the various
elements of the negligence inquiry. Stapleton has explained that the question of the extent of
responsibility is distinct from the question of causal responsibility, and different terminology
applies to each issue. Where damage is divisible and the defendant caused only a portion of the
total damage, he is only causally responsible for that portion, so may only be held liable for that
portion. Stapleton says:
Though it is a rough usage it is convenient to call this causal ‘apportionment’ to
distinguish it from the notion of ‘proportionate’ liability which refers to the, rarely
adopted, rule that a tortfeasor is only liable to the claimant for a proportion of an
indivisible injury as well as distinguishing it from the notion of ‘contribution’ between
tortfeasors.105
Beever (n38) 488. See for example Holtby v Brigham & Cowan (Hull) Ltd [2000] 3 All ER 421 (CA), discussed at Ch
2 (n75) and (n94).
105 Stapleton, ‘Lords a’leaping ‘ (n15) 283.
104
210
In contrast, in cases of indivisible damage, liability cannot be ‘apportioned’ on the basis of causal
responsibility. Instead liability must be divided on the basis of non-causal considerations of
comparative responsibility, and this is properly called ‘several’ liability. This means that if the
damage in Barker really was the risk of mesothelioma, which is a divisible harm, the defendant
has only caused a distinct portion of the overall risk exposure. Since he is causally responsible for
only a portion of the total risk, legal liability can only arise in respect of this portion of the total
risk, so there is no need to then apply several liability. As Beever explains, ‘[t]he defendant
created the whole of the relevant risk. It does not make sense to speak of attributing any share of
that risk to him’.106 The fact that a kind of ‘apportionment’ was achieved through several liability
shows that it is still the mesothelioma itself that forms the gist of the action.107 This is also
evident in the fact that the ‘risk’ is valued as a portion of the outcome harm rather than having
intrinsic value of its own.108
So far it has been shown that requiring the claimant to have developed mesothelioma before
allowing him to bring a claim means that the gist of the negligence action has not been redefined
as the risk of mesothelioma. The focus will now turn to the question of whether risk could form
the gist of a negligence action if the physical harm requirement was to be abandoned.
3.3 Can risk constitute damage in corrective justice?
This section focuses on risk solely in the forward-looking sense that the term properly implies.
The question is whether exposing somebody to the risk of mesothelioma or of other harm is
capable of constituting damage within a corrective justice-based system of negligence law. In this
forward-looking sense it is still important to draw a distinction similar to that in the previous
chapter on loss of a chance. In that chapter it was seen that causal processes can be deterministic
Beever (n38) 488.
Note that the Supreme Court in Sienkiewicz (n18), and the majority in BAI v Durham (n4), confirmed that the gist
of the action is the mesothelioma.
108 The question of the value of a risk will be considered below.
106
107
211
or indeterministic.109 It is only in indeterministic processes that we can appropriately say that
there is a ‘chance’ of an outcome occurring and that the probability of this outcome is ‘objective’.
In deterministic processes, as most physical processes are believed to be, there is not a ‘chance’
as such since the occurrence of a particular outcome can theoretically be predicted with certainty
given limitless knowledge. Instead the probability used to describe the likelihood of the outcome
occurring is an epistemological probability, it reflects our belief based on the limited knowledge
that is available. Thus a patient who is unwell when he visits his doctor is either destined to
recover, or destined to succumb, he does not have an objective chance of recovery. Generally
speaking then we deal with epistemological and not objective probabilities.
Recalling that probability is one element of risk, along with the possible outcome, risk can be
objective or subjective even though it is based on epistemological probability. The term ‘risk’
denotes uncertainty as to the likelihood of a future outcome, so it incorporates epistemological
probability. The harmful agent/risk agent has the potential to cause an outcome i.e. it has the
potential to form part of a deterministic causal process, but the limits of our knowledge prevent
us from saying whether it will or will not cause harm to a particular individual. But there is a limit
to this because the relevant outcome has an objective nature. With an indivisible harm such as
mesothelioma, one only creates an objective risk of mesothelioma if the potential victim has not
already contracted the disease. Of course, an employer will not usually know whether the
employee has already contracted mesothelioma, so from his perspective exposing the employee
to asbestos exposes the employee to a risk of mesothelioma, so in this subjective/epistemic
sense the conduct creates a risk of harm. So even though risk relies on epistemological probability,
the objective nature of harm means that we can talk about objective risk and
subjective/epistemic risk. The following sections will consider whether risk, in either sense, can
constitute damage in the negligence inquiry.
109
See Ch 3 section 4.
212
3.3.1 Objective risk: the evidentiary gap prevents proof that the defendant created
an objective risk
A criticism of the ‘risk as damage’ approach that has emerged in academic writing is that the
evidentiary gap also prevents the claimant from proving that the defendant’s negligence
increased the risk that he would develop mesothelioma.110 Scherpe has explained:
The aetiology of the disease is such that once contracted further exposure does not
matter and certainly cannot increase the risk of contracting it. If the disease was
contracted during the first employment, all following exposure did and could not
increase the risk and hence there could not be a contribution to the risk. But whether
that was the case, we do not know – which is exactly the point: we do not know whether
all exposure actually increased the risk. Assuming that it did is therefore also resorting to
a fiction.111
In other words, it is only before mesothelioma has been contracted that exposure to asbestos
actually creates an objective risk that the individual will develop the disease. Since the evidentiary
gap prevents us from saying when the disease was contracted it is also impossible to say that any
exposures other than the first exposure actually created a risk of the disease for this individual.
Nolan has addressed a potential criticism of this argument:
It could be argued that this is to adopt the wrong perspective. From the point of view of
the defendant at the time of the breach of duty, a risk was imposed on the claimant…and
indeed this must always be the case where the defendant has been negligent towards the
claimant, as he has to be in order to be liable. But even if we assume that this was the
perspective the House of Lords had in mind when they said that in cases falling within
the Fairchild principle, the defendant was deemed to have caused the injury when he
could be shown to have materially increased the risk of its happening, this switch from
objective risk to subjective risk cannot rescue the risk as damage idea. This is because
110
111
See Nolan (n72) 178; Beever (n38) 486; Scherpe (n89) 488.
Scherpe (n89) 488.
213
even if we accept the premise that an unrealised risk is a form of harm, this claim can
surely be sustainable only if the defendant’s action did in fact create a risk for the
claimant.112
Since the evidentiary gap prevents us from being able to say whether the defendant exposed the
claimant to an objective risk of mesothelioma, risk in the objective sense cannot help overcome
the problem of the evidentiary gap.
3.3.2 Subjective/epistemic risk: lacks the moral significance to be damage
The only remaining sense in which risk may therefore constitute damage, if it is to assist the
courts in overcoming the evidentiary gap, is in the subjective sense. As explained above, a lack of
knowledge will often prevent an employer from knowing whether his employee has already
developed mesothelioma, so exposing the employee to asbestos exposes him to a risk of
mesothelioma in a subjective, or epistemic sense. This is what enables the conduct to be
characterised as wrongful, or negligent.113 Since the standard of care in negligence is an objective
standard, for the avoidance of confusion the label ‘epistemic’ will be used instead of ‘subjective’.
As noted above, Nolan argued that while the Fairchild principle may be based on the fact that
from the defendant’s perspective he exposed the claimant to the risk of mesothelioma this does
not rescue risk as a form of damage. It will be argued that Nolan’s assertion is correct and that
epistemic risk cannot constitute damage in negligence. One barrier is the difficulty of explaining
why this sort of risk matters and should be deserving of compensation. The other, far greater,
barrier is that it adds nothing to the negligence inquiry. Corrective justice requires a wrongdoer
to repair the loss that he caused the victim to suffer through his wrongdoing. If the concept of
‘loss’ is the same as the concept of wrongdoing, i.e. the creation of an unreasonable risk, then
loss is effectively subsumed into the concept of wrongdoing and liability is imposed for the sole
fact of wrongdoing towards the claimant. An entire aspect of the equation has been removed,
112
113
Nolan (n72) 179.
Clearly asbestos exposure is also wrongful because it creates a risk of other asbestos-related illnesses too.
214
and the focus is solely on the defendant’s wrongdoing so the justification for liability can no
longer be corrective justice but must be a defendant-focused form of justice such as retributive
or distributive justice.
3.3.2.1 The difficulty of explaining why epistemic risk deserves compensation
In terms of the negligence doctrines, Stapleton has argued that the question of what constitutes
‘gist damage’ or actionable damage is ‘rarely addressed squarely by courts’, and that the word
‘damage’ is ‘bandied about in a number of different contexts, usually without clear definition yet
equally without apparent awareness of the importance of precision in its use’.114 This highlights
one of the key difficulties with accepting risk as damage in cases involving the evidentiary gap –
should exposure to the risk of harm be considered damage for the purposes of corrective justicebased liability? This question was not addressed head-on by the House of Lords. Amirthalingam
has therefore argued that ‘Barker is an unsatisfactory decision in that it does not explain why
“increased risk” should qualify as the gist of negligence’.115 Furthermore, it does not account for
why the risk of mesothelioma, and possibly of other diseases whose aetiology involves the same
kind of evidentiary gap, is considered to be damage but the risk of other physical harm is not
regarded as damage. It will be argued that any limit on what kinds of risk qualify as damage will
necessarily be arbitrary because of the more fundamental problem that the risk of harm simply
cannot constitute damage. Each of these objections must be addressed in greater detail.
A significant barrier to accepting that the risk of harm constitutes damage is that it is difficult to
say why risk has value and what that value is. Voyiakis explains that ‘we cannot decide whether
persons exposed to risk of physical harm should be entitled to claim compensation unless we
have some rough idea of what they would demand to be compensated for’.116 As discussed in the
Jane Stapleton, ‘The Gist of Negligence: Part 1 Minimal Actionable Damage’ (1988) 104 LQR 213, 213.
Kumaralingam Amirthalingam, ‘The Changing Face of the Gist of Negligence’ in J Neyers, E Chamberlain & S
Pitel (eds.) Emerging Issues in Tort Law (Hart Publishing 2007) 474.
116 Emmanuel Voyiakis, ‘The Great Illusion: Tort Law and Exposure to Danger of Physical Harm’ (2009) 72 MLR
909, 915.
114
115
215
previous chapter, the risk itself does not have a value as a proportion of the value of the physical
harm to which it relates:
[W]hen I make it 40 per cent more likely that you will suffer lung cancer, I am not
causing you 40 per cent of the harm that lung cancer brings about. The quantity of risk
and the quantity of physical harm do not seem to be related in any such straightforward
way. We therefore have reason to object to the latter being taken as a measure of the
former.117
Instead, Voyiakis focuses on the concrete effects that exposure to the risk of harm can create,
notably preventative medical care and the increase in the cost of health insurance or life
insurance. Steele has distinguished the ‘utility value’ of risk from its ‘mathematical value’ and
explained that the utility value of any particular risk depends on the individual involved. She says
that ‘if the possible losses are not purely financial, it is much harder to attach the risk ‘value’ in
any calculable form at all’.118 So even the question of whether a risk has a financial impact will
depend on the individual and whether they are inclined to have health/life insurance anyway.
Similarly, risk cannot be valued according to the anxiety it causes the claimant. The House of
Lords held in Rothwell that the risk of future asbestos-related illness is not actionable damage, nor
is anxiety related to this risk, nor is the combination of the two.119 Green supports this
conclusion, noting that human beings necessarily operate within a notion of risk meaning that:
the degree to which any individual might or might not actually be at risk of an adverse
outcome need bear no resemblance whatsoever to his perception of that risk. Therefore,
any anxiety founded upon such a necessarily epistemic conception of that risk is a step
further removed from the defendant’s breach of duty than is anxiety consequent upon
actual harm resulting from the tort.120
ibid 917.
Steele (n91) 25.
119 Rothwell (n87).
120 Sarah Green, ‘Risk Exposure and Negligence’ (2006) 122 LQR 386, 388.
117
118
216
It is also difficult to reconcile risk as damage with the decision in Cartledge v E Jopling & Sons
Ltd.121 The claimant there had suffered damage to his lungs caused by inhalation of silica dust
and it was held that the fibres caused actionable damage before producing symptoms that were
noticeable in everyday life. If physical damage is actionable before it is discoverable by the
claimant then risk exposure ought to also be actionable regardless of whether the claimant is
aware of the risk. Yet risk exposure only seems to acquire significance if the claimant is aware of
it. If the claimant is unaware of the risk then he is unaffected unless it materialises in physical
harm.
The argument that epistemic risk should be regarded as damage could be seen to be analogous
with the argument that was made in the previous chapter that the loss of an epistemic chance
should be regarded as damage in the limited context of medical negligence involving the
misdiagnosis or mistreatment of existing illness. That argument, however, was based on the
unique characteristic of the doctor’s duty of care which is focused on the illness, and the fact that
both the patient and doctor regard the patient as having a ‘chance’ of recovery. There is no such
focus in this context. Amirthalingam has noted, therefore, that limiting recovery for risk to
mesothelioma, or to risks arising in the course of employment, would be arbitrary – there is
nothing marking these contexts out as ones in which the claimant and defendant regard the risk
of harm as damage in advance of their interaction. Amirthalingam suggests that ‘unacceptable
risk’ might be considered damage. He explains that this ‘provides a normative framework for
courts to determine what sort of risks should be classed as unacceptable, and incrementally
develop the law’.122 This solution is too vague to be applied in practice, and it still does not
explain why a particular risk is unacceptable.
121
122
[1963] AC 758 (HL).
Amirthalingam (n115) 475.
217
3.3.2.2 Epistemic risk: the conflation of breach and damage
The more fundamental obstacle to the recognition of epistemic risk as damage in negligence is
that the damage requirement would be subsumed into the breach inquiry so liability would be
based entirely on wrongdoing rather than on the interaction of the two parties, so it would no
longer be a system of corrective justice.
In Fairchild, Lord Rodger suggested that ‘[a]t best, it was only good luck if any particular
defendant’s negligence did not trigger the victim’s mesothelioma’.123 This is true, but should not
impact on corrective justice-based liability since this addresses wrongful loss. As discussed in
Chapter One,124 the causation requirement has been criticised as introducing an element of
‘moral luck’ into negligence liability,125 and if the focus of liability was the wrongdoing of the
defendant alone then there may be some force to this argument. However since the focus of
corrective justice is not on either of the parties alone, but is on the inequality that occurs when
the wrongdoing of one person causes a loss to another person, it is only concerned with equality
in transactions. When two people interact there is only a transaction when something passes
between them i.e. one causes a loss to the other. As Steele has explained, ‘if the defendant is
lucky and does not cause damage, then potential victims are lucky too and do not suffer any’. 126
The claimants in these cases have not been ‘lucky’ overall because we know that they have
developed mesothelioma but this does not mean that it was caused by the particular defendant.
If the defendant was lucky and did not trigger the claimant’s mesothelioma but simply created a
risk of mesothelioma by exposing him to asbestos dust then there has been no corrective justice
transaction between the defendant and the claimant.
The relationship between the Fairchild test and breach of duty was considered by the Court of
Appeal in Williams v University of Birmingham.127 The main issue to be decided in that case was
Fairchild (n13) [155] (Lord Rodger).
See Ch 1 text to (n88).
125 David Howarth, ‘Many duties of care – or a duty of care? Notes from the underground’ (2006) 26 OJLS 449, 461.
126 Steele (n91) 116.
127 Williams (n101).
123
124
218
whether the defendant had breached his duty of care towards the claimant. The judge at first
instance said that the legal standard of care was ‘to take all reasonable measures to ensure that
[the claimant] was not exposed to a material increase in the risk of mesothelioma’.128
In the Court of Appeal, however, it was held that this was inaccurate:
A reference to exposure “to a material increase in the risk of mesothelioma” brings the test for
causation in mesothelioma cases into the prior questions of the nature of the duty and
what constitutes a breach of it. There is nothing in Fairchild or [Sienkiewicz v Greif] to
suggest that the House of Lords or the Supreme Court has altered the “breach of duty”
test in mesothelioma cases so that a claimant only has to demonstrate that the defendant
failed to take reasonable steps to ensure that the claimant or victim was not exposed to a
“material increase in the risk of mesothelioma”.129
Instead the duty of care is ‘to take reasonable care…to ensure that [the claimant] was not
exposed to a foreseeable risk of asbestos related injury’.130 If the defendant did not materially
increase the risk of harm then it was clear that he had not breached his duty of care. But if his
conduct did materially increase the risk of harm then the court must still ask whether a
reasonable person in the defendant’s place would have taken further steps to reduce the risk of
harm. The test of breach of duty is therefore more demanding than the test of material increase
in the risk of harm that is applied at the causation stage, so it is clear that epistemic risk is already
subsumed into the breach inquiry. Subsuming the damage requirement into the breach inquiry
would mean that the only remaining elements of the negligence inquiry would be the duty of care
and the breach of duty. The negligence action would focus almost exclusively on the defendant,
and would no longer be a system of corrective justice.
(Belcher J) at first instance (unreported) cited in Williams (CA) (n101) [39]
Williams (CA) (n101) [40] (Aikens LJ).
130 ibid [40].
128
129
219
4. Theoretical approaches to the ‘evidentiary gap’
In this chapter it has been argued that the ‘evidentiary gap’ presents an insurmountable obstacle
to proof of causation. This cannot coherently be overcome by taking a more relaxed evidential
approach and drawing a common sense inference of causation. Nor can it be avoided by moving
the problem to a different aspect of negligence by reconceptualising the gist of the action. This
final section therefore turns to the question of how to justify imposing liability, whether it is
several or joint and several, when causation cannot be proven. The dominant theme in judicial
reasoning in Fairchild and in Barker is that the solutions were justified by policy concerns and the
demands of fairness and justice. It will be argued here that ‘fairness’ and ‘justice’ are not
arguments of corrective justice. The first part seeks to clear away a number of possible
misconceptions about this. In the second part it will be demonstrated that they reflect an
instrumentalist approach to liability, although the exact content of these instrumentalist goals is
not clearly articulated in the decisions. As a result of adopting an instrumentalist approach, the
Fairchild principle lacks a coherent theoretical foundation since, as discussed in Chapter One,
instrumentalist theories cannot be properly implemented through the institution of negligence
law. As Nolan has explained, ‘[t]he trouble is that the logic of the instrumentalist approach has
no obvious stopping point…And because there are no logical stopping points – or rather
because the logical stopping points are incompatible with the basic idea of tort law - the courts
have had to resort to illogical ones instead’.131
4.1 Dispelling possible misconceptions
First it is important to address a number of possible misconceptions that ‘fairness’ may have
some claim on corrective justice.
131
Nolan (n72) 189.
220
4.1.1 McGhee: the defendant definitely caused the loss
In Fairchild, Lord Bingham noted that the Court of Appeal had sought to distinguish that case,
where multiple former employers may have caused the disease, from McGhee where there was
only one employer involved. He said that in McGhee, ‘there was a risk that the defendant might
be held liable for acts for which he should not be held legally liable but no risk that he would be
held liable for damage which (whether legally liable or not) he had not caused’. 132 This, however,
is not a corrective justice-based reason for imposing liability. In Chapter One it was seen that a
similar argument arose in relation to Chester v Afshar.133 There the defendant doctor had
negligently failed to warn the patient about a particular risk involved in an operation, and the risk
did in fact materialise even though the operation itself was carried out without negligence. The
patient may still have undergone the operation if she had been properly warned, so she would
still have been exposed to the risk that materialised, which meant that the failure to warn her of
the risk was not a cause of her loss. The defendant doctor was held liable despite the absence of
the causal link. In order to achieve this outcome, however, the House of Lords had to rely on
distributive arguments such as the importance of vindicating the patient’s right to give informed
consent, and the need to deter doctors from breaching their duty to warn patients of risks. 134 In
corrective justice, a person is only required to repair loss that is caused by their wrongdoing. If loss
was caused by non-wrongful, i.e. non-negligent, conduct then it is not a wrongful loss. It is
therefore irrelevant that the defendant employer in McGhee was responsible for all of the
claimant’s exposure to brick dust. Some of this exposure was negligent but some was innocent
and unless it can be shown that the negligent exposure specifically was a cause of the claimant’s
dermatitis then the loss is not wrongful.
Fairchild (n13) [33] (Lord Bingham). See also Joe Thomson, ‘Barker v Corus: Fairchild chickens come home to
roost’ (2006) Edin LR 421, 424.
133 [2004] UKHL 41; [2005] 1 AC 134. See Ch 1 text to (n85).
134 See Ch 1 section 1.5.1
132
221
4.1.2 Fairchild: the ‘innocent claimant’ argument
In Fairchild, the House of Lords considered that any background risk of mesothelioma was so
small it could be effectively discounted, so that it was possible to say that the claimant’s harm
had definitely been caused by the negligence of a former employer it was just impossible to say
which former employer(s) in particular.135 Lord Bingham said:
[T]here is a strong policy argument in favour of compensating those who have suffered
grave harm, at the expense of their employers who owed them a duty to protect them
against that very harm and failed to do so, when the harm can only have been caused by
breach of that duty.136
Assuming for now that any non-negligent exposure could properly be disregarded, this would
mean that the claimant could prove that he had suffered a ‘wrongful loss’ because he could
prove that his loss had been caused by somebody else’s negligence, so the claimant’s side of the
corrective justice equation would be fulfilled. This is similar to the problem that arose in the
American case of Sindell v Abbott Laboratories.137 The claimants were young women who had
developed a cancer which was caused by their mothers having taken the drug Diethylstilbestrol
(DES) during pregnancy. They were unable to prove which particular manufacturer had
produced the drug that their mother had taken. Goldberg has said that ‘factual causation was not
disputed in the DES cases, but what became an issue was the identification of the particular
manufacturer of the product and a causal link between it and the damage’.138 But as Beever
explains:
We cannot say that the claimant is deserving of recovery and leave it at that. It is
irrelevant to corrective justice that the claimant deserves recovery from someone…In
Fairchild (n13) [2] (Lord Bingham).
Fairchild (n13) [33] (Lord Bingham). See also [41] (Lord Nicholls), [153] (Lord Rodger).
137 26 Cal. 3d 588, 607 P 2d 924 (1980).
138 Richard Goldberg, Causation and Risk in the Law of Torts: Scientific Evidence and Medicinal Product Liability (Hart
Publishing 1999) 11.
135
136
222
corrective justice, it is necessary to say that the claimant is deserving of recovery from some
particular person.139
This is because corrective justice is not concerned with the claimant and defendant individually,
but with both of them in their interaction, so it is not enough to say that the claimant has
suffered a wrongful loss. He must be able to say that his wrongful loss was caused by the
particular defendant’s wrongdoing before that particular defendant can have any corrective
justice-based interpersonal responsibility to repair the loss. It is important that the responsibility is
individual as highlighted by Nolan. He argues that corrective justice cannot accommodate the
focus on the claimant’s wrongful loss because this ‘refers not to each defendant as an individual
but to the defendants as a collective entity’.140 He explains:
[W]hile it seems reasonable to say that, as between the claimant and the defendants, the
equities favour the claimant, unless there is some good reason why we should
‘collectivise’ the defendants in this way, we still do not have a justification for the
imposition of liability as between the claimant and each individual defendant.141
Moreover, ‘from the individual defendant’s point of view, it seems irrelevant that the other
possible cause of the claimant’s injury is the negligence of another person, as opposed to nonnegligent conduct or a natural event’.142 Even if the claimant has suffered a wrongful loss, in the
sense that his loss was definitely caused by somebody else’s negligence, corrective justice still
only requires that loss to be repaired by the individual whose negligence caused it. There may be
a strong reason to feel that the claimant is deserving of compensation, but if a particular
defendant cannot be proved to have caused the loss then there is no corresponding reason to
engage his personal responsibility.
The above discussion is premised on the assumption in Fairchild that all the sources of asbestos
had been created by the negligence of former employers. This is not always the case. In Barker
Beever, Rediscovering (n38) 457.
Nolan (n72) 174.
141 ibid 174.
142 ibid 175.
139
140
223
the claimant had negligently exposed himself to asbestos for a period of time. More importantly,
it seems that the court was wrong to ignore the background risk through environmental
exposure. The majority of the claimant’s exposure to asbestos in Sienkiewicz was not due to
negligence but was present in the atmosphere of the area where she lived. Mesothelioma can be
caused by very fleeting exposures to asbestos.143 Nolan suggests that imposing liability on a
defendant in these circumstances reflects an underlying assumption ‘that somehow fault matters
more than causation’.144 But, as he explains:
[W]ithin a corrective justice framework this assumption is simply false, for, as Allan
Beever points out, ‘from the perspective of corrective justice, the defendant is entirely
innocent (with respect to the claimant) unless the claimant establishes all [the elements of
the cause of action] against the defendant’.145
Imposing liability, whether joint and several or merely several, on a defendant whose negligence
has not been proven to be a cause of the claimant’s loss is simply not consistent with corrective
justice-based interpersonal responsibility. Indeed, although the attempt to reformulate the gist of
the action as the risk of harm in Barker was ultimately flawed, it did highlight the House of
Lords’ awareness that the causation requirement is an indispensable element of the negligence
inquiry. Commenting on a similar approach in American tort law, Ripstein and Zipursky explain:
Some recent cases sidestep causation issues by redefining the category of injury that will
permit recovery…All of these efforts tend to underscore, rather than to rebut, the
centrality of causation in extant American tort law, as they indicate the need to shift to
other categories, rather than abandoning causation itself.146
While it may be argued that there is a clear need to compensate the claimants in these cases, and
that it is important to deter defendants from acting negligently, it is not the role of negligence
Stapleton, ‘Lords a’leaping’ (n15) 276-9.
Nolan (n72) 175.
145 ibid 175, citing Beever, Rediscovering (n38) 446.
146 Arthur Ripstein and Benjamin C Zipursky, ‘Corrective Justice in an Age of Mass Torts’ in Gerald J Postema (ed.)
Philosophy and the Law of Torts (Cambridge University Press 2001) 219.
143
144
224
liability to pursue these goals. The coherence of negligence law is dependent on its pursuit of
corrective justice-based personal responsibility, so it cannot be adapted to plug the gaps in order
to achieve these consequentialist goals. As seen in Chapter One, if the law is ‘stretched’ to
achieve compensation and/or deterrence then it will suffer in terms of coherence and tend to be
measured in terms of how well it achieves these consequentialist aims leading critics, such as
Atiyah, to suggest that negligence law should be abandoned in favour of a fairer compensation
system.147 It is important to recall that while negligence law gives effect to corrective justice this
does not preclude distributive goals such as compensation and deterrence from being pursued
through other, more appropriate, institutions. As Ripstein and Zipursky argue:
[A]ll of the considerations put forward as reasons for attenuating the causation
requirement in tort law are more plausibly viewed as reasons for supplementing the legal
system with public compensatory and deterrent systems, for tort law without causation
will be both ineffective and arbitrary to the core.148
In the absence of state schemes it may seem unjust to leave the claimant without compensation
in negligence, and to allow the negligent defendant to escape liability, but this is unjust from a
distributive perspective and from the perspective of corrective justice liability must be denied. As
explained in Chapter One, it is one thing to acknowledge that negligence liability has
compensatory and deterrent effects, but it is another thing to say that it has compensatory and
deterrent goals and to measure its success against those goals. As a system of compensation or
deterrence it has been shown that negligence will necessarily be a failure because the goals are
mutually truncating and incompatible with the correlative structure of the negligence action.149 If
we deny liability where causation is not proven then negligence law might be criticised for failing
to achieve optimal compensation and deterrence but it is not appropriate to measure its success
or failure on those terms. It would be like measuring a car in terms of how good it is as an
See Ch 1 text to (n57) and (n58), P S Atiyah, The Damages Lottery (Hart Publishing 1997) 32-95.
Ripstein and Zipursky (n146) 245.
149 See Ch 1 text to (n54) and (n55).
147
148
225
aeroplane and criticising it because it is no good at flying – we are criticising it for failing to be
something that it is not. Instead we are interested in how well negligence law succeeds at
achieving corrective justice and, measured in those terms, negligence law is a success if it denies
liability where causation cannot be proven.
4.2 Where next?
This final section turns to the question of how the courts should resolve future cases that involve
an evidentiary gap. Since the passing of the Compensation Act 2006, these cases must be divided
into two categories: those cases involving mesothelioma, and those involving other diseases.
4.2.1 Mesothelioma
The House of Lords decision to impose several liability in Barker was reversed, and joint and
several liability was reinstated in cases involving mesothelioma by s.3 Compensation Act 2006.
The provisions of s.3 Compensation Act 2006 apply where the defendant is liable in tort in
connection with damage caused to the victim by the disease ‘whether by reason of having
materially increased a risk or for any other reason’.150 As Lord Phillips noted in Sienkiewicz, as and
when more is understood about mesothelioma and the uncertainty surrounding the aetiology is
removed the Fairchild test may no longer apply since there would be no need for an exceptional
approach to causation if the evidentiary gap no longer existed.151 It is therefore essential that the
courts assess future developments in the scientific evidence relating to the aetiology of
mesothelioma. This chapter has shown the importance of using the NESS test to understand
precisely which aspect of causation is affected by the evidentiary gap. It has also examined the
relationship between scientific and legal standards of proof, and highlighted the need for judges
to engage fully with scientific evidence so that they understand not only the limits to what it can
prove but also the full extent of what it is capable of proving. If science progresses to a stage
150
151
s3(1) Compensation Act 2006.
Sienkiewicz (n18) [70] (Lord Phillips).
226
where it is able to describe a sufficient set of conditions to cause mesothelioma then the
evidentiary gap will no longer exist. Any remaining difficulties of proof will relate to the necessity
of the defendant’s negligence for the sufficiency of an actually occurring set of conditions. This
may be easier in some cases than others, but it will at least be notionally possible. Given that the
Fairchild test is inconsistent with corrective justice, the courts should cease to apply it as soon as
the evidentiary gap is closed.
4.2.2 Other diseases involving an evidentiary gap
The Compensation Act 2006 only applies to mesothelioma which means that if the Fairchild test
is applied to other diseases involving an evidentiary gap the Barker principle of several liability is
applicable. Although Lord Brown in Sienkiewicz suggested that the Fairchild exception should be
limited to mesothelioma actions,152 it has not been convincingly restricted in this way so it may
potentially be applied more widely.153 This final section considers whether the Fairchild exception
can and should apply to claims involving any other diseases where there is an evidentiary gap. It
will be argued that the ‘single harmful agent’ rule adopted in Wilsher and approved in Fairchild
provides a rational limit on when the Fairchild exception can apply. However, turning to the
question of whether it should be applied outside the context of mesothelioma, it will be argued
that since the Fairchild exception lacks a coherent theoretical justification it ought not to be
applied to other diseases. Although the application of several liability following Barker is a ‘fairer’
solution than joint and several liability, this does not rescue the Fairchild exception from its lack
of a theoretical basis.
152
153
ibid [187]
Chris Miller, ‘Causation in personal injury after (and before) Sienkiewicz’ (2012) 32 LS 396, 410.
227
4.2.2.1 The single harmful agent rule: Wilsher v Essex Area Health Authority154
As discussed in chapter two, the claimant in Wilsher was a baby born prematurely who later
suffered blindness caused by retrolentalfibroplasia (‘RLF’) and his case was brought in the
interval between McGhee and Fairchild. To recap the facts of this case, the defendant’s negligence
consisted of administering excess oxygen which carries a risk of RLF and the baby also suffered
four other conditions due to being born prematurely, each of which also carried a risk of RLF.
He was unable to prove on the balance of probabilities that the excess oxygen was a cause of his
RLF using the but-for test. He therefore argued instead that the McGhee principle should be
applied, and the defendant should be held liable on the basis that his negligence materially
increased the risk of RLF. The majority of the Court of Appeal found that the McGhee principle
was applicable because the aetiology of the claimant’s RLF was unknown, and applying the
McGhee test they found that the defendant’s negligence had materially increased the risk of RLF.
Browne-Wilkinson V-C dissented, and his dissenting opinion was approved in the House of
Lords. He distinguished the facts in Wilsher because there was only one harmful agent in McGhee
i.e. brick dust, whereas in Wilsher there were five possible harmful agents i.e. the excess oxygen
created by the defendant and the four other naturally occurring conditions.
The defendants failed to take reasonable precautions to prevent one of the possible
causative agents (e.g. excess oxygen) from causing RLF. But no one can tell in this case
whether excess oxygen did or did not cause or contribute to the RLF suffered by the
plaintiff. The plaintiff’s RLF may have been caused by some completely different agent
or agents… To the extent that certain members of the House of Lords [in McGhee]
decided the question on inferences from evidence or presumptions, I do not consider
that the present case falls within their reasoning. A failure to take preventative measures
154
[1988] AC 1074 (HL).
228
against one out of five possible causes is no evidence as to which of those five caused the
injury.155
Since it could not be proved that excess oxygen was a cause of the claimant’s RLF it was not
possible to say that the defendant’s negligence had led to a material increase in an operative risk:
[U]nless and until you can say that the plaintiff’s RLF was caused by oxygen, it is
impossible to say that the injury falls “squarely within the risk”.156
The House of Lords reversed the decision of the Court of Appeal in Wilsher, preferring BrowneWilkinson’s ‘single harmful agent’ rule.157 Although Lord Hoffmann dissented on this point in
Fairchild, arguing that the rule does not draw a ‘principled distinction’,158 he later accepted a more
nuanced form of the single harmful agent rule in Barker explaining:
In my opinion it is an essential condition for the operation of the exception that the
impossibility of proving that the defendant caused the damage arises out of the existence
of another potential causative agent which operated in the same way. It may have been
different in some causally irrelevant respect…but the mechanism by which it caused the
damage…must have been the same. So, for example, I do not think that the exception
applies when the claimant suffers lung cancer which may have been caused by exposure
to asbestos or some other carcinogenic matter but may also have been caused by
smoking and it cannot be proved which is more likely to have been the causative agent.159
The single harmful agent rule has been widely criticised. Nolan, for example, says that ‘it is
difficult to see any logic in this limitation’,160 and Beever says that it ‘seems to be nothing more
Wilsher v Essex AHA [1987] 1 QB 730 (CA), 779 (Browne-Wilkinson VC).
ibid 780 (Browne-Wilkinson VC).
157 It will be recalled that Lord Bridge held that McGhee had not laid down a new legal principle but that the court
had taken a ‘robust and pragmatic approach’ to the facts in order to draw a legitimate inference of causation. The
House of Lords in Fairchild subsequently overruled this aspect of Lord Bridge’s reasoning, holding that McGhee had
in fact laid down a new legal principle and that the inferential approach was inappropriate, but it affirmed the
outcome in Wilsher, so the ‘single harmful agent’ rule still applies.
158 Fairchild (n13) [72].
159 Barker (n73) [24] (Lord Hoffmann).
160 Nolan (n72) 173.
155
156
229
than an arbitrary restriction on the application of Fairchild’.161 It will be argued that from a purely
policy-driven perspective the rule is indeed arbitrary, but from a factual causation perspective it is
justified because it limits the level of scientific uncertainty under which a defendant may still be
held liable.
As this chapter has shown, the decision in Fairchild is inconsistent with the principles of
corrective justice, and instead the House of Lords explicitly resorted to distributive justice-based
arguments to justify their decision. Distributive arguments were also raised in Fairchild to
rationalise the single harmful agent rule that had been established in Wilsher. Unlike McGhee and
Fairchild which involved an employer’s negligence, Lord Hoffmann said that in the medical
context of Wilsher ‘it cannot possibly be said that the duty to take reasonable care in treating
patients would be virtually drained of content’ unless the rules on causation are relaxed. 162
Furthermore, ‘the political and economic arguments involved in the massive increase in the
liability of the National Health Service…are far more complicated than the reasons…for
imposing liability upon an employer who has failed to take simple precautions’. 163 Viewed in
these terms the single harmful agent rule is indeed arbitrary. If liability is imposed because of the
need to enforce the defendant’s duty of care,164 or because of the need to compensate the
claimant,165 then it is irrelevant whether there was a single harmful agent involved.
The policy arguments about liability are, however, not the appropriate starting point for our
reasoning. The Fairchild exception addresses a problem of proof of factual causation so it is
essential to begin by identifying the exact nature of the causal problem before considering the
appropriate solution. In terms of the NESS test for causation, it has been explained above that
where there is an evidentiary gap it is impossible to describe a ‘sufficient set’ in general terms.
Beever, Rediscovering (n38) 475.
Fairchild (n13) [69] (Lord Hoffmann).
163 ibid [69].
164 ibid [33] (Lord Bingham).
165 ibid [36] (Lord Nicholls).
161
162
230
But it is important here to specify precisely why this was impossible and also to restate the terms
of the NESS test:
[A] particular condition was a cause of (condition contributing to) a specific consequence
if and only if it was a necessary element of a set of antecedent actual conditions that was
sufficient for the occurrence of the consequence.166
In both McGhee and Fairchild it was known that the sufficient set that actually occurred contained
the type of substance to which the defendant had exposed the claimant. The sufficient set that
actually occurred in McGhee contained abrasions from brick dust, and in Fairchild it contained
inhalation of asbestos fibres. What was unknown was how the particular substance forms a
sufficient set – does a sufficient set contain just one or many abrasions/asbestos fibres? In
Wilsher, there was an extra degree of uncertainty because it was impossible to say that the
sufficient set that actually occurred even contained excess oxygen. A sufficient set can contain
excess oxygen. A sufficient set might also contain excess oxygen in combination with one or
more of the other conditions.167
If the baby in Wilsher had not suffered the other conditions then it might be said that an
evidentiary gap remains in relation to oxygen since Mustill LJ explained in the Court of Appeal
that ‘[w]hat is no longer so clear as it was, is that the concentration rather than the duration of
the exposure is crucial’.168 It is not possible to say how oxygen causes RLF since there is
uncertainty what ‘excess’ means. It may relate to the concentration of oxygen, or to the duration,
or both. The court could thus apply the Fairchild test and say that a doctor who negligently
exposed a baby to very high levels of oxygen for a short time/a slight excess for a long time, had
increased the risk of RLF. But there is an extra level of uncertainty in the actual facts of Wilsher
because it is not even possible to say that oxygen was part of the sufficient set that actually occurred.
Given that it was very exceptional for the court to depart from corrective justice and allow the
Wright, ‘Causation in Tort Law’ (n26) 1790 (emphasis added).
Wilsher (CA) (n155) 764-5 (Mustill LJ).
168 ibid 766.
166
167
231
claimant to leap the evidentiary gap in Fairchild, it would be even more exceptional and require
even stronger justification to allow a claimant to leap an even wider evidentiary gap such as
existed in Wilsher.
4.2.2.2 Limiting the Fairchild exception to mesothelioma
The decision in Fairchild was quite openly motivated by considerations of ‘fairness’ and ‘justice’
and other ‘policy’ considerations. Similarly in Barker the legal rules were adapted to achieve what
was perceived as a fair solution, with Lord Walker going so far as to say ‘I prefer to start with the
more fundamental issue of apportionment, since it must have a bearing on how far and how fast
the boundaries of the new principle are to be extended’.169 This openness has received some
praise, with Morgan commenting that the ‘open acknowledgment [in Fairchild] that policy had to
be considered was welcome’,170 and Khoury suggesting that in Barker:
The House of Lords has arguably been more transparent and honest than the French
Cour de cassation by recognising that proportional liability is, in such cases, imposed
without proof of causation between the negligence and the actual loss, due, essentially, not to
logical conceptual reasons, but to the existence of overriding policy objectives.171
But it is not sufficient merely to acknowledge openly that a decision departs from traditional
principles because of policy concerns. In order to be justifiable, those policy concerns must be
articulated and defended. Morgan thus criticised the Fairchild decision because ‘[t]he court must
explain precisely which policies are at stake, and such analysis is missing in Fairchild’.172 Broad
reference to ‘fairness’ and ‘justice’ is also insufficient as Morgan explains:
Barker (n73) [107] (Lord Walker). See also [40]-[43] (Lord Hoffmann), [101] (Lord Rodger), [109] Lord Walker,
[127] (Baroness Hale).
170 Jonathan Morgan ‘Lost Causes in the House of Lords: Fairchild v Glenhaven Funeral Services’ (2003) MLR 277, 279.
171 Lara Khoury, ‘Causation and Risk in the Highest Courts of Canada, England and France (2008) 124 LQR 103,
130.
172 Morgan (n170) 279.
169
232
While these are noble sentiments, they should not be dignified with the label of policy.
Such a ‘policy’ says nothing more than that ‘injured claimants should recover’, and
therefore it is much too wide to be of any use in setting the boundaries of what, after all,
is said to be an exceptional approach…A convincing justification must explain why
recovery is allowed in certain cases, and in those cases only.173
The argument that it is unjust to leave claimants uncompensated is also criticised by Nolan who
says ‘this position was not really fully articulated, no justification being given for the assertion
that irredeemable evidential uncertainty should operate to the detriment of the negligent
defendants rather than the blameless employee’.174 It is also difficult to reconcile with the fact
that under Barker there is a significant chance that the claimant will not be fully compensated if
defendants can no longer be traced or have become insolvent. Kramer argues that ‘while it is
clearly unfortunate that the victims are not fully compensated, arguably this is a result of
tortfeasors becoming insolvent and not of the apportionment rule itself’.175 But if the Fairchild
exception was devised to achieve a fair outcome for claimants in the face of a practical problem
of proof, surely this entire area of law is shaped by the desire to achieve a particular outcome in
practice, so it does not make sense to ignore the practical impact of insolvency of defendants
when designing the legal rule. As the decision in BAI v Durham attests, the decision in Barker has
potential to impact on the interpretation of the terms of insurance contracts relating to the date
that the injury was ‘sustained’ or ‘contracted’ because this may depend on whether the relevant
damage is the risk or the physical harm.176
It should be remembered that s.3 Compensation Act 2006 only applies to mesothelioma, so the
Barker principle of apportionment would apply to other evidentiary gap disease cases. The
decision to apply several liability in Barker seems fairer than joint and several liability, but what is
fairness being measured against? Green has argued that ‘although there may well exist strong
ibid.
Nolan (n72) 172.
175 Adam Kramer, ‘Smoothing the Rough Justice of the Fairchild Principle’ (2006) 122 LQR 547, 550.
176 BAI v Durham (n4).
173
174
233
arguments in favour of maintaining the rule against apportionment where causation can be
established on an orthodox basis, it is perfectly reasonable to argue for a modification of that
rule where the causal criteria have themselves been altered’.177 This shows the significance of the
Barker decision – the abandonment of the causation requirement in Fairchild meant that liability
was no longer based on corrective justice which meant that other negligence doctrines could also
be altered to achieve a particular vision of liability. As such, Steel explains ‘[t]he real function of
‘risk’ in Barker is as a practical tool for devising apportionment of liability in the interests of
fairness’.178 Steel goes on to explain that ‘[s]uch a consideration of fairness is only intelligible
against the background that there is a (moral) difficulty with the proposition that a defendant
who has not caused the mesothelioma may be liable for it’.179
One proposal for providing a rational limit to Fairchild is that it could be limited to cases of
‘industrial risk’, but it will be argued that this does not provide a sound theoretical basis for the
rule. As noted above, Lord Hoffmann considered that it was appropriate to apply the Fairchild
principle in a claim brought by an employee against a former employer but not one brought by a
patient against their doctor. Indeed, Amirthalingam has argued that to avoid throwing causation
‘into the abyss of intuition and policy…it would have been better if the case had been analysed
squarely as one about liability for industrial (or unacceptable) risk’.180 The difficulty with
characterising the Fairchild exception as involving industrial risk is that this focuses on defendants
to the detriment of the claimants. As discussed in Chapter One, deterrence is a distributive goal
centred on defendants and it cannot be properly achieved through the institution of negligence
law which involves claimants too. Given that liability in Fairchild, and under s.3 Compensation
Act 2006, is joint and several, defendants are arguably being over-deterred. If deterrence is to be
implemented through the negligence system which requires a claimant to bring an action then
Sarah Green ‘Winner takes all’ (2004) 120 LQR 566, 568.
Sandy Steel, ‘Uncertainty Over Causal Uncertainty: Karen Sienkiewicz (Administratrix of the Estate of Enid Costello
Deceased) v Greif (UK) Ltd’ (2010) 73 MLR 646, 653. See also Barker (n73) [120]-[128] (Baroness Hale).
179 Steel (n178) 653.
180 Amirthalingam (n115) 470-1.
177
178
234
there ought to be a focus on the needs of the claimant too. But if Fairchild was a principle relating
only to industrial risk then a claimant exposed to asbestos in school (or university) would be
unable to bring a successful claim against the school or the workmen employed by the school. 181
And although Amirthalingam’s focus on the industrial context seems to emphasise the deterrent
effect of liability, he also seems concerned by the vulnerability of the claimant in the industrial
context because he favours joint and several liability (subject to reduction for contributory
negligence) since this ‘leaves the risk of insolvency of other potential defendants on the
defendants’.182 As Chapter One demonstrated, the structure of negligence law prevents
distributive goals from being achieved in a comprehensive way, and within a negligence claim the
competing distributive arguments relate to the claimant and the defendant separately and are
mutually truncating.
The difficulty with this is that ‘fairness’ is still not explained, beyond saying that some concession
is made to both the claimant and to the defendant. Once fairness is prioritised over corrective
justice in developing legal rules, the courts ought to consider what is fair to both claimants and
defendants. This pluralism introduces incoherence because, as Chapter One showed, goals
relating to the claimant and defendant individually will be mutually truncating.183 This has been
seen in other areas of negligence. In Gwilliam v West Herts NHS Trust,184 the court imposed a duty
of care on the defendant to take reasonable care to ensure that a contractor had public liability
insurance. This duty did not arise because of considerations of interpersonal responsibility
between the claimant and defendant but because of the court’s desire to ensure that
compensation could be paid. Morgan has criticised this decision because insurance is a
distributive institution so while it may be desirable for it to operate alongside the negligence
system it cannot drive the content of the legal principles. As he argues, ‘[t]he courts are
Such as occurred in Knowsley Metropolitan Borough Council v Willmore [2011] UKSC 10, [2011] 2 WLR 523, and
Williams v University of Birmingham (n101).
182 Amirthalingam (n115) 474.
183 See Ch 1, section 1.3
184 Gwilliam v West Herts NHS Trust [2002] EWCA Civ 1041; [2003] QB 443.
181
235
powerless to complete the change to a (rational) compensation scheme…Thus, they must ensure
that tort continues to function in its own terms, namely the legal doctrinal argument which
characterises private law’.185 The same criticism applies to Fairchild.
Given that negligence law cannot fully achieve goals relating to either the claimant or the
defendant, the decision about where the balance between the competing interests ought to lie is
subjective. Morgan has highlighted this by contrasting the decisions of the Court of Appeal and
the House of Lords in Fairchild:
The perceived unfairness of [leaving the claimant without compensation] had…not
swayed the Court of Appeal…who set against the unpalatability of denying recovery to
sympathetic claimants the injustice of requiring a defendant who may well have had no
causal connexion with the relevant injury to compensate the claimant’s entire loss…The
House of Lords were content to assert that this injustice was ‘plainly’ outweighed by the
injustice of the claimants recovering nothing, which reasoning is (equally plainly)
contestable and certainly not much of an explanation.186
With this in mind, the decision in Barker to apply several liability seems to be the fairest solution
because it seeks to find a balance between the competing interests of the claimant and the
defendant. But it must be made clear that this is a personal preference as to which distributive
arguments ought to be prioritised over others in a context where such a choice simply has to be
made and articulated clearly because of the underlying incompatibility of achieving distributive
goals through a system of corrective justice. Given that proportionate liability cannot be justified
by principles of corrective justice-based interpersonal responsibility it is preferable for the
Fairchild exception to be restricted to claims in respect of mesothelioma.
185
186
Jonathan Morgan, ‘Tort, insurance and incoherence’ (2004) 67 MLR 384, 400.
Jonathan Morgan, ‘Lost Causes’ (n170) 281.
236
Conclusion
This chapter has shown that it is important to understand the different approaches in Barker in
detail. On first impression, Lord Hoffmann’s reconceptualisation of the gist of the action as the
risk of harm appears to reconcile the Fairchild test with the principles of corrective justice; since
the claimant can only prove that the defendant’s negligence exposed him to the risk of harm he
is only compensated for the risk exposure rather than the physical harm. The subtle change in
the formulation of the Fairchild test also appears to bring it in line with the Wardlaw test of
material contribution to harm, the only difference being that the ‘harm’ where Fairchild/Barker
applies is ‘the risk of harm’. In comparison, the reasoning adopted by Baroness Hale seems
flimsy and unprincipled since it is based on vague notions of what she feels is a ‘fair’ balance
between the competing interests of the claimant and defendant. This chapter has shown,
however, that Lord Hoffmann’s approach is conceptually problematic and is not consistent with
corrective justice. This means that just like the Fairchild exception, the Barker apportionment rule
is motivated by considerations of distributive justice and instrumentalism. Given that such
considerations cannot provide a coherent foundation for negligence liability, Baroness Hale’s
open acknowledgment that the solution was based on vague considerations of fairness more
accurately reflects the inherent uncertainty created by instrumentalist accounts of negligence.
The developments of the Fairchild test therefore serve to illustrate the incoherence that is
inevitable when courts adapt the rules of causation to pursue instrumentalist goals rather than
corrective justice. Since causation is an essential ingredient of corrective justice-based personal
responsibility, there should be no liability when the causal link cannot be proved. While this may
leave a pressing social need to compensate victims of a particular disease, as well as a need to
deter defendants from engaging in risky conduct, it is not the primary function of negligence law
to pursue these objectives. Compensation, deterrence and other instrumentalist goals do not
provide a coherent theoretical justification for negligence, so the success of negligence law
should not be measured against how well it achieves these goals. As noted in Chapter One,
237
Weinrib has observed that ‘the legal regime of personal injuries can be organised either
correctively or distributively…nothing about personal injury as such consigns it to the domain of
a particular form of justice’.187 In light of the evidentiary gap relating to mesothelioma which
prevents attribution of interpersonal responsibility in most cases this is a particular type of
personal injury that ought to be addressed through a distributive mechanism.
187
Ernest Weinrib, ‘Corrective Justice’ (1992) 77 Iowa L Rev 403, 415.
238
Conclusion
This thesis set out to identify a coherent legal solution to the problem of the ‘evidentiary gap’ in
causation in negligence but, as explained in the introduction, it was first necessary to situate the
issue within the broader context of causation in negligence more generally. This involved
addressing both the nature and function of the tort of negligence as well as the role played by
causation within that tort.
In Chapter One, it was argued that the tort of negligence is most coherently conceived as a
system of corrective justice. It was seen that Aristotelian corrective justice and Kantian Right
together form an account of interpersonal responsibility that prioritises coherence and morality.
Causation was shown to be an essential element of interpersonal responsibility, but it was also
seen that it has a limited role; causation provides the factual nexus between the claimant and the
defendant but it must be supplemented by notions of wrongdoing and damage which sculpt the
boundaries of interpersonal responsibility. This must be reflected in the negligence doctrines, so
while the various doctrines must join together to ‘articulate a single normative sequence’, the
doctrine of causation must have a clearly defined role within that sequence. Chapter One
therefore explained the interrelationship between the causation requirement, and the elements of
actionable damage, duty of care, breach of duty and quantification of loss. It was also argued that
the duty of care is the central concept within the negligence inquiry because it is at the duty stage
that the issues of interpersonal responsibility in the relationship between the claimant and
defendant are identified, and these issues guide the application of the remaining negligence
doctrines in each case.
Having shown in Chapter One that causation has a vital but limited role in interpersonal
responsibility, Chapter Two considered the demands that negligence law makes of the causation
doctrine and argued that it is essential to isolate factual causation, i.e. the fact of being a cause,
from evaluative conclusions that a condition was the responsible cause. As a test of factual
causation, the NESS test was shown to be preferable to the but-for test because it is better
239
matched to the philosophical account of what it means to be a cause. In straightforward cases
the NESS test does not complicate the causal inquiry, and in more complex cases it is able to
resolve causal problems that the but-for test cannot. It was also shown that when applying the
NESS test it is necessary to clearly identify the damage forming the gist of the action from the
outset because the causal problem will vary depending on whether the damage is ‘divisible’ or
‘indivisible’. Applying the NESS test, twinned with a clear understanding of damage, it was
possible to clarify the scope and meaning of the Wardlaw test of ‘material contribution to harm’.
This analysis showed that in cases of divisible damage where there is more than one source of
the harmful agent the Wardlaw test equates to an application of the but-for test to a portion of
the total damage suffered by the claimant. In cases of indivisible damage the Wardlaw test may
constitute an exception to the but-for test. Crucially, however, it does not constitute an
exception to the requirement of causation; it merely compensates for the conceptual
inadequacies of the but-for test. The NESS test operates in a simpler way to overcome the same
conceptual inadequacies
Having defined clearly the role of causation, and the conceptual requirements for a test of
causation, the thesis then turned to evidential problems relating to proof of causation. Chapter
Three considered the issue of loss of a chance in medical negligence cases. This argument was
raised in Hotson and in Gregg by claimants who were unable to prove that the defendant doctor’s
negligent delay in diagnosis of their illness had caused them to suffer physical harm. Instead they
sought proportionate recovery of the physical harm by reformulating the damage as the loss of a
chance of avoiding physical harm. Drawing on the understanding of the interrelationship of the
doctrines of damage, causation, and quantification previously set out in Chapter One, it was
argued that the House of Lords was right to reject their claims because a patient in these
circumstances does not have an objective chance of being cured. This means that if
proportionate recovery were allowed it would effectively award damages for the physical harm
but apply a discount to reflect the degree of doubt surrounding proof of causation. However,
240
this chapter proposed a more limited role for loss of a chance as an independent form of damage
leading to limited recovery in some medical negligence scenarios. It was argued that although an
epistemological chance should not normally be considered to constitute damage, it becomes a
meaningful loss in this specific context because of the issues of interpersonal responsibility
between the doctor and patient in the treatment of existing illness.
Chapter Four turned to the issue of the evidentiary gap that first arose in McGhee and
subsequently in Fairchild and, drawing on the wider understanding of negligence and causation
developed in the previous chapters, asked how to achieve a coherent solution to this problem.
The problem of the evidentiary gap relates to the scientific understanding of the aetiology of the
disease, so the NESS test is unable to resolve the problem but it does enable us to pinpoint the
precise nature of the causal problem facing the claimant. This highlighted that it is not possible
to define a ‘sufficient set’ of conditions to cause the disease, so it is not possible to draw a
meaningful inference of causation. The McGhee/Fairchild test of ‘material increase in the risk of
harm’ clearly therefore constitutes a relaxation of the causation requirement, so it makes a
departure from principles of corrective justice.
This chapter evaluated Lord Hoffmann’s attempt to rationalise the Fairchild test in Barker by
reconceptualising the gist of the action as the ‘risk of harm’ and apportioning liability according
to the extent of the defendant’s negligent contribution to the total risk. It was shown that this
solution is not consistent with corrective justice, because it conflates the notions of damage and
breach by making ‘risk exposure’ central to both.
Both the Fairchild test and the Barker apportionment principle were shown to be inconsistent
with principles of corrective justice and motivated instead by considerations of distributive
justice such as the desire to compensate the victims of mesothelioma in Fairchild, and the desire
to achieve a ‘fair’ balance between the burdens placed on each party in Barker. As shown at the
outset of this thesis, distributive goals cannot provide a coherent foundation for negligence law;
the bipolar structure of negligence liability imposes an artificial limit on the attainment of
241
distributive goals, and they are mutually truncating. This results in indeterminacy and
incoherence in the law. Where the Fairchild principle applies it is preferable that it should apply in
conjunction with the Barker apportionment principle in order to balance the competing interests
of claimants and defendants, but this rests on little more than the author’s personal preference
as, inevitably, any solution premised on distributive principles must. It is therefore important that
the application of the Fairchild test should be limited to claims for mesothelioma. It is also
important that the courts continue to follow advances in medical science and that they revert to
orthodox principles of causation as and when the evidentiary gap is closed.
242
Bibliography
Aagard TS, ‘Identifying and Valuing the Injury in Lost Chance Cases’ (1998) 96 Mich L Rev
1335
Amirthalingam, K ‘The Changing Face of the Gist of Negligence’ in Neyers, Chamberlain and
Pitel (eds) Emerging Issues in Tort Law (Hart Publishing 2007)
Anderson GM, ‘Disease Causation and the Extent of Material Contribution’ (2006) 15 SLT 87
Atiyah PS, The Damages Lottery (Hart Publishing 1997)
Bagshaw R, ‘Publication Review: Responsibility and Fault’ (2000) 116 LQR 321
Bailey SH, ‘Causation in Negligence: What is a Material Contribution?’ (2010) 30 LS 167
Barnes DW, ‘Too Many Probabilities: Statistical Evidence of Tort Causation’ (2001) 64 LCP 191
Beauchamp TL & Rosenberg A, Hume and the Problem of Causation (Oxford University Press 1981)
Beever A, ‘Cause-in-fact: Two Steps Out of the Mire’ (2001) 51 UTLJ 327
— — ‘Gregg v Scott and Loss of a Chance’ (2005) 24 UQLJ 201
— — Rediscovering the Law of Negligence (Hart Publishing 2007)
— — ‘Corrective Justice and Personal Responsibility in Tort Law’ (2008) 28 OJLS 475
— — ‘Formalism in Music and Law’ (2011) 61 UTLJ 213
Bevan N, ‘Williams v University of Birmingham: Personal Injury – Negligence – Asbestos’ [2012] JPI
Law 1
Boon A, ‘Causation and the Increase of Risk’ (1988) 51 MLR 506
Bradford Hill A, ‘The environment and disease: association or causation?’ (1965) 58 Proc R Soc
Med 295
Broadbent A, ‘Fact and Law in the Causal Inquiry’ (2009) 15 Legal Theory 173
Broadie S & Rowe C, Aristotle: Nicomachean Ethics: Translation, Introduction and Commentary (Oxford
University Press 2002)
Brown S and Lindsey S, ‘Seeing the Wood for the Trees – Loftus-Brigham and Apportionment of
Damage’ (2005) 21 Const LJ 431
i
Burrows A, ‘Uncertainty About Uncertainty: Damages for Loss of a Chance’ [2008] JPI Law 31
Cane P, ‘Justice and Justifications for Tort Liability’ (1982) 2 OJLS 30
— — ‘Corrective Justice and Correlativity in Private Law’ (1996) 16 OJLS 471
— — The Anatomy of Tort Law (Hart Publishing 1997)
— — ‘Distributive Justice and Tort Law’ [2001] NZ L Rev 401
— — ‘Responsibility and Fault: a Relational and Functional Approach to Responsibility’ in Peter
Cane and John Gardner (eds) Relating to Responsibility: Essays in Honour of Tony Honoré on his 80th
Birthday (Hart Publishing 2001)
— — ‘Tort Law as Regulation’ (2002) 31 CLWR 305
— — Responsibility in Law and Morality (Hart Publishing 2002)
— — ‘The Anatomy of Private Law Theory: A 25th Anniversary Essay’ (2005) 25 OJLS 203
— — ‘Rights in Private Law’ in Donal Nolan and Andrew Robertson (eds) Rights and Private Law
(Hart Publishing 2011)
Caun L, ‘Multiple Causes of Injury’ [2003] JPI Law 96
Chapman B, ‘Ernie’s Three Worlds’ (2011) 61 UTLJ 179
Coggon D and Taylor AN, ‘Causation and Attribution of Disease in Personal Injury Cases: A
Scientific Perspective’ [2009] JPI Law 12
Coleman J, ‘Moral Theories of Torts: Their Scope and Limits: Part 1’ (1982) 1 Law and Phil 371
— — ‘Corrective Justice and Wrongful Gain’ (1982) 11 JLS 421
— — ‘Moral Theories of Torts: Their Scope and Limits: Part 2’ (1983) 2 Law and Phil 5
— — Risks and Wrongs (Oxford University Press 1992)
— — ‘The Mixed Conception of Corrective Justice’ (1992) 77 Iowa L Rev 427
Davies M, ‘The Road From Morocco: Polemis Through Donoghue to No-Fault’ (1982) 45 MLR
534
Englard I, Corrective and Distributive Justice: From Aristotle to Modern Times (Oxford University Press
2009)
Finkelstein C, ‘Is Risk a Harm?’ (2003) 151 U Pa L Rev 963
Fischer DA, ‘Tort Recovery for Loss of a Chance’ (2001) 36 Wake Forest L Rev 605
— — ‘Insufficient Causes’ (2006) 94 Ky LJ 277
ii
Fletcher GP, ‘Law and Morality: A Kantian Perspective’ (1987) 87 Colum L Rev 533
Fumerton R and Kress K, ‘Causation and the Law: Preemption, Lawful Sufficiency, and Causal
Sufficiency’ (2001) 64 LCP 83
Gardner J, ‘What is Tort Law For? Part 1: The Place of Corrective Justice’ (2011) 30 Law & Phil
1
Gold S, ‘Causation in Toxic Torts: Burdens of Proof, Standards of Persuasion, and Statistical
Evidence’ (1986) 96 Yale LJ 376
Goldberg R, Causation and Risk in the Law of Torts: Scientific Evidence and Medicinal Product Liability
(Hart Publishing 1999)
— — ‘Causation, idiopathic conditions and the limits of epidemiology’ (2009) 13 Edin LR 282
Gordley J, ‘Tort Law in the Aristotelian Tradition’ in David G Owen (ed) Philosophical Foundations
of Tort Law (Clarendon Press 1995)
Green L, ‘The Causal Relation Issue in Negligence Law’ (1962) 60 Mich L Rev 543
Green S, ‘Winner Takes All’ (2004) 120 LQR 566
— — ‘The Risk Pricing Principle: A Pragmatic Approach to Causation and Apportionment of
Damages’ (2005) 4 LP&R 159
— — ‘Coherence of Medical Negligence Cases: A Game of Doctors and Purses’ (2006) 14 Med
LR 1
— — ‘Risk Exposure and Negligence’ (2006) 122 LQR 386
— — ‘Contributing to the Risk of Confusion? Causation in the Court of Appeal’ (2009) 125
LQR 44
Grubb A, ‘Causation and Medical Negligence’ [1988] 47 CLJ 350
Gullifer L, ‘One Cause After Another’ (2001) 117 LQR 403
Hamer D, ‘‘Chance Would be a Fine Thing’: Proof of Causation and Quantum in an
Unpredictable World’ (1999) 23 Melb U L Rev 557
Hart HLA and Honoré T, Causation in the Law (2nd ed., Clarendon Press 1985)
Hill T, ‘A Lost Chance for Compensation in the Tort of Negligence by the House of Lords’
(1991) 54 MLR 511
iii
Hoffmann L, ‘Causation’ (2005) 121 LQR 592
Hogg M, ‘Re-establishing Orthodoxy in the Realm of Causation’ (2007) 11 Edin L R 8
— — ‘Causation and Apportionment of Damages in Cases of Divisible Injury’ (2008) 12 Edin
LR 99
— — ‘Duties of Care, Causation, and the Implications of Chester v Afshar’ (2005) 9 Edin LR 156
— — ‘Developing Causal Doctrine’ in Richard Goldberg (ed) Perspectives on Causation (Hart
Publishing 2011)
Honoré T, ‘The Morality of Tort Law – Questions and Answers’ in David G Owen (ed)
Philosophical Foundations of Tort Law (Oxford University Press 1995)
— — ‘Necessary and Sufficient Conditions in Tort Law’ in David G Owen (ed.) Philosophical
Foundations of Tort Law (Oxford University Press 1995)
— — ‘Medical non-disclosure, causation and risk: Chappel v Hart’ (1999) 7 TLJ 1
— — Responsibility and Fault (Hart Publishing 1999)
Howarth D, ‘Three Forms of Responsibility: On the Relationship Between Tort Law and the
Welfare State’ (2001) 60 CLJ 553
— — ‘Many Duties of Care – Or a Duty of Care? Notes from the Underground’ (2006) 26 OJLS
449
Hume, An Abstract of a Treatise of Human Nature, ed. JM Keynes and P Sraffa (Cambridge
University Press 1938) 11f.
Jansen N, ‘The Idea of a Lost Chance’ (1999) 19 OJLS 271
— — ‘Duties and Rights in Negligence: A Comparative and Historical Perspective on the
European Law of Extracontractual Liability’ (2004) 24 OJLS 443
Jones MA, ‘Proving Causation – Beyond the “But For” Test’ (2006) 22 PN 251
— — ‘Causation and Psychiatric Damage’ (2008) 24 PN 255
Judicial College, Guidelines for the Assessment of General Damages in Personal Injury Cases (11th ed.,
Oxford University Press 2012)
Kant I, Foundations of the Metaphysics of Morals, (trans. Lewis White Beck) (Bobbs-Merrill Education
Publishing 1969)
Keeler J, ‘Thinking Through the Unthinkable: Collective Responsibilities in Personal Injury Law’
(2001) 30 CLWR 349
iv
Kelley PJ, ‘Causation and Justice: A Comment’ (1978) 56 Wash U L Q 635
Khoury L, Uncertain Causation in Medical Liability (Hart Publishing 2006)
— — ‘Causation and Risk in the Highest Courts of Canada, England and France’ (2008) 124
LQR 103
King Jr. JH, ‘Causation, Valuation, and Chance in Personal Injury Torts Involving Preexisting
Conditions and Future Consequences’ (1981) 90 Yale LJ 1353
Kinsky C, ‘SAAMCo 10 years on: Causation and Scope of Duty in Professional Negligence
Cases’ (2006) 22 PN 86
Klimchuk D, ‘On the Autonomy of Corrective Justice’ (2003) 23 OJLS 49
Kramer A, ‘Smoothing the Rough Justice of the Fairchild Principle’ (2006) 122 LQR 547
Kress K, ‘Formalism, Corrective Justice and Tort Law’ (1991) 77 Iowa L Rev i
Laleng P, ‘Sienkiewicz v Greif (UK) Ltd and Willmore v Knowsley Metropolitan Borough Council: A
Material Contribution to Uncertainty?’ (2011) 74 MLR 767
Lee J, ‘Causation in Negligence: Another Fine Mess’ (2008) 24 PN 194
Lewis D, ‘Causation’ (1973) 70 J Phil 556
Lord Hoffmann, ‘Causation’ in Richard Goldberg (ed.) Perspectives on Causation (Hart Publishing
2011)
Lord Hope of Craighead, ‘James McGhee – A Second Mrs Donoghue?’ (2003) 62 CLJ 587
Lord Neuberger of Abbotsbury, ‘Loss of a Chance and Causation’ (2008) 24 PN 206
Lunney M, ‘What Price a Chance?’ (1995) 15 LS 1
MacCormick N, ‘Taking Responsibility Seriously’ (2005) 9 Edin LR 168
Mackie JL, The Cement of the Universe: A Study of Causation (Clarendon Press 1980)
Martin R, ‘Categories of Negligence and Duties of Care: Caparo in the House of Lords’ (1990) 53
MLR 824
Maskrey S and Edis W, ‘Chester v Afshar and Gregg v Scott: Mixed Messages for Lawyers’ [2005] JPI
Law 205
v
McBride N, ‘Duties of Care – Do They Really Exist?’ (2004) 24 OJLS 417
McGregor H, ‘Loss of Chance: Where has it Come From and Where is it Going?’ (2008) 24 PN
21
McIvor C, Third Party Liability in Tort (Hart Publishing 2006)
— — ‘Debunking some judicial myths about epidemiology and its relevance to personal injury
litigation’ (unpublished conference paper, Obligations VI: Challenging Orthodoxy conference,
Canada 18th-20th July 2012)
— — ‘The Use of Epidemiological Evidence in UK Tort Law’ in S Loue (ed.) Global Perspectives
in Forensic Epidemiology (forthcoming)
Miller C, ‘Coal Dust, Causation and Common Sense’ (2000) 63 MLR 763
— — ‘Judicial approaches to contested causation: Fairchild v. Glenhaven Funeral Services in
context’ (2002) 1 LP&R 119
— — ‘Gregg v Scott: Loss of Chance Revisited’ (2005) 4 LP&R 227
— — ‘Causation in Personal Injury: Legal or Epidemiological Common Sense?’ (2006) 26 LS
544
— — ‘Loss of chance in personal injury: a review of recent developments’ (2006) 5 LP&R 63
— — ‘Liability for Negligently Increased Risk: the Repercussions of Barker v Corus (UK) Plc’
(2009) 8 LP&R 39
— — ‘Causation in personal injury after (and before) Sienkiewicz’ (2012) 32 LS 396
Moore MS, Causation and Responsibility: An Essay in Law, Morals, and Metaphysics (Oxford University
Press 2009)
Morgan J, ‘Lost Causes in the House of Lords: Fairchild v Glenhaven Funeral Services’ (2003) 66
MLR 277
— — ‘Tort, Insurance and Incoherence’ (2004) 67 MLR 384
— — ‘The Rise and Fall of the General Duty of Care’ (2006) 22 PN 206
— — ‘Policy Reasoning in Tort Law: the Courts, the Law Commission and Critics (2009) 125
LQR 215
Mullany NJ, ‘Common Sense Causation – An Australian View’ (1992) 12 OJLS 431
Mullender R, ‘Negligence Law, the Welfare State, and “Our Moral Life”’ (2009) 25 PN 187
Neyers JW, ‘The Inconsistencies of Aristotle’s Theory of Corrective Justice’ (1998) 11 CJLJ 311
vi
Noah L, ‘An Inventory of Mathematical Blunders in Applying the Loss-of-a-Chance Doctrine’
(2005) 24 Rev Litig 369
Nolan D, ‘Causation and the Goals of Tort Law’ in Andrew Robertson and Tang Hang Wu
(eds.) The Goals of Private Law (Hart Publishing 2009)
Oppenheim R, ‘The “Mosaic” of Tort Law: the Duty of Care Question’ [2003] JPI Law 151
Owen DG, ‘Why Philosophy Matters to Tort Law’ in David G Owen (ed) Philosophical
Foundations of Tort Law (Clarendon Press 1995)
Peel E, ‘Lost Chances and Proportionate Recovery: Barker v Corus’ [2006] LMCLQ 289
Perry SR, ‘The Moral Foundations of Tort Law’ (1992) 77 Iowa L Rev 449
— — ‘Professor Weinrib’s Formalism: The Not-So-Empty Sepulchre’ (1993) 16 Harv J L & Pub
Pol’y 597
— — ‘Risk, Harm, and Responsibility’ in David G. Owen (ed.) Philosophical Foundations of Tort
Law (Oxford University Press 1995)
— — ‘Honoré on Responsibility for Outcomes’ in Peter Cane and John Gardner (eds) Relating to
Responsibility: Essays in Honour of Tony Honoré on his 80th Birthday (Hart Publishing 2001)
— — ‘Responsibility for Outcomes, Risk, and the Law of Torts’ in Gerald J Postema (ed)
Philosophy and the Law of Torts (Cambridge University Press 2001)
— — ‘The Role of Duty of Care in a Rights-Based Theory of Negligence Law’ in Andrew
Robertson and Tang Hang Wu (eds) The Goals of Private Law (Hart Publishing 2009)
Plowden S and Volpe H, ‘Fairchild and Barker in MRSA Cases: Do Fairchild and Barker Provide an
Argument for a Relaxation of Causation Principles in Claims for Hospital Acquired MRSA?’
[2006] JPI Law 259
Porat A and Stein A, Tort Liability Under Uncertainty (Oxford University Press 2002)
— — ‘Indeterminate Causation and Apportionment of Damages: An Essay on Holtby, Allen, and
Fairchild’ (2003) 23 OJLS 667
Posner RA, ‘The Concept of Corrective Justice in Recent Theories of Tort Law’ (1981) 10 JLS
187
Price DPT, ‘Causation: The Lords’ Lost Chance?’ (1989) 38 ICLQ 735
Reece H, ‘Losses of Chances in the Law’ (1996) 59 MLR 188
vii
Ripstein A and Zipursky BC, ‘Corrective Justice in an Age of Mass Torts’ in Gerald J Postema
(ed) Philosophy and the Law of Torts (Cambridge University Press 2001)
Robertson A, ‘Constraints on Policy-Based Reasoning in Private Law’ in Andrew Robertson and
Tang Hang Wu (eds) The Goals of Private Law (Hart Publishing 2009)
— — ‘Justice, Community Welfare and the Duty of Care’ (2011) 127 LQR 370
— — ‘Rights, Pluralism, and the Duty of Care’ in Donal Nolan and Andrew Robertson (eds)
Rights and Private Law (Hart Publishing 2011)
— — ‘On the Function of the Law of Negligence’ (unpublished conference paper, Obligations
VI: Challenging Orthodoxy conference, Canada 18th-20th July 2012)
Robertson DW, ‘Causation in the Restatement (Third) of Torts: Three Arguable Mistakes (2009)
33 Wake Forest L Rev 1007
Robinson GO, ‘Probabilistic Causation and Compensation for Tortious Risk’ (1985) 14 JLS 779
Scherpe J, ‘A New Gist?’ (2006) 65 CLJ 487
Schroeder CH, ‘Corrective Justice and Liability for Increasing Risks’ (1989) 37 UCLA L Rev 439
— — ‘Causation, Compensation, and Moral Responsibility’ in David G Owen (ed) Philosophical
Foundations of Tort Law (Clarendon Press 1995)
Scott W, ‘Causation in Medico-Legal Practice: A Doctor’s Approach to the ‘Lost Opportunity’
Cases’ (1992) 55 MLR 521
Sheinman H, ‘Tort Law and Corrective Justice’ (2003) 22 Law & Phil 21
Smith J, ‘Causation – the Search for Principle’ [2009] JPI Law 101
Smith SA, ‘Publication Review: Idea of Private Law’ (1996) 112 LQR 363
— — Contract Theory (Clarendon Press 2004)
Stanton K, ‘Professional Negligence: Duty of Care Methodology in the Twenty First Century’
(2006) 22 PN 134
Stapleton J, ‘The Gist of Negligence: Part 1 Minimal Actionable Damage’ (1988) 104 LQR 213
— — ‘The Gist of Negligence: Part 2 The Relationship Between “Damage” and Causation’
(1988) 104 LQR 389
— — ‘Duty of Care and Economic Loss: A Wider Agenda’ (1991) 107 LQR 249
— — ‘Tort, Insurance and Ideology’ (1995) 58 MLR 820
viii
— — ‘Negligent Valuers and Falls in the Property Market’ (1997) 113 LQR 1
— — ‘Duty of Care Factors: a Selection from the Judicial Menus’ in Peter Cane and Jane
Stapleton (eds) The Law of Obligations: Essays in Celebration of John Fleming (Oxford University Press
1998)
— — ‘Publication Review – Causation and Risk in the Law of Torts: Scientific Evidence and
Medicinal Product Liability’ (2000) 116 LQR 506
— — ‘Unpacking “Causation”’ in Peter Cane and John Gardner (eds) Relating to Responsibility:
Essays in Honour of Tony Honoré on his 80th Birthday (Hart Publishing 2001)
— — ‘Lords a'leaping evidentiary gaps’ (2002) 10 TLJ 276
— — ‘Cause in Fact and the Scope of Liability for Consequences’ (2003) 119 LQR 388
— — ‘The Golden Thread as the Heart of Tort Law: Protection of the Vulnerable’ (2003) 24
Aust Bar Rev 135
— — ‘Loss of the Chance of Cure from Cancer’ (2005) 68 MLR 996
— — ‘Two Causal Fictions at the Heart of U.S. Asbestos Doctrine’ (2006) 122 LQR 189
— — ‘Occam’s Razor Reveals an Orthodox Basis for Chester v Afshar’ (2006) 122 LQR 426
— — ‘Choosing What We Mean by “Causation” in the Law’ (2008) 73 Mo L Rev 433
— — ‘Factual Causation and Asbestos Cancers’ (2010) 126 LQR 351
— — ‘The Fairchild Doctrine: Arguments on Breach and Materiality’ (2012) 71 CLJ 32
Stauch M, ‘Causation, Risk, and Loss of Chance in Medical Negligence’ (1997) 17 OJLS 205
— — ‘Taking the Consequences for Failure to Warn of Medical Risks’ (2000) 63 MLR 261
— — ‘Risk and Remoteness of Damage in Negligence’ (2001) 64 MLR 191
— — The Law of Medical Negligence in England and Germany (Hart Publishing 2007)
— — ‘“Material contribution” as a response to causal uncertainty: time for a rethink’ (2009) 68
CLJ 27
Steel S, ‘Uncertainty Over Causal Uncertainty: Karen Sienkiewicz (Administratrix of the Estate of Enid
Costello Deceased) v Greif (UK) Ltd’ (2010) 73 MLR 646
— — and Ibbetson D, ‘More Grief on Uncertain Causation in Tort’ (2011) 70 CLJ 451
Steele J, Risks and Legal Theory (Hart Publishing 2004)
— — Tort Law: Texts, Cases and Materials (2nd ed., Oxford University Press 2011)
Stevens R, Torts and Rights (Oxford University Press 2007)
Stiggelbout M, ‘The Case of ‘Losses in Any Event’: a Question of Duty, Cause or Damages?’
(2010) 30 LS 558
ix
Thomson J, ‘The Raising of Lazarus: The Resurrection of McGhee v National Coal Board’ (2003) 7
Edin LR 80
— — ‘Barker v Corus: Fairchild Chickens Come Home to Roost’ (2006) 10 Edin LR 421
Tse MH, ‘Tests for Factual Causation: Unravelling the Mystery of Material Contribution,
Contribution to Risk, the Robust and Pragmatic Approach and the Inference of Causation’
(2008) 16 TLJ 249
Uberoi M, ‘Does the Law of Tort Still Recognise The Concept of Individual Responsibility?’
(2003) 7 EPIS 1
Voyiakis E, ‘The Great Illusion: Tort Law and Exposure to Danger of Physical Harm’ (2009) 72
MLR 909
Waldron J, ‘Moments of Carelessness and Massive Loss’ in David G Owen (ed.) Philosophical
Foundations of Tort Law (Oxford University Press 1995)
Walker VR, ‘Uncertainties in Tort Liability for Uncertainty’ (2002) 1 LP&R 175
Weekes R, ‘Not Seeing the Wood for the Trees – Risk Analysis as an Alternative to Factual
Causation in Fairchild’ (2003) 12 Nott LJ 18
Weinrib EJ, ‘A Step Forward in Factual Causation’ (1975) 38 MLR 518
— — ‘Toward a Moral Theory of Negligence Law’ (1983) 2 Law & Phil 37
— — ‘The Insurance Justification and Private Law’ (1985) 14 JLS 681
— — ‘Law as a Kantian Idea of Reason’ (1987) 87 Colum L Rev 472
— — ‘Legal Formalism: On the Immanent Rationality of Law’ (1988) 97 Yale LJ 949
— — ‘The Special Morality of Tort Law’ (1989) 34 McGill L J 403
— — ‘Corrective Justice’ (1992) 77 Iowa L Rev 403
— — ‘The Jurisprudence of Legal Formalism’ (1993) 16 Harv J L & Pub Pol’y 583
— — ‘The Gains and Losses of Corrective Justice’ (1994) 44 Duke LJ 277
— — The Idea of Private Law (Harvard University Press 1995)
— — ‘Corrective Justice in a Nutshell’ (2002) 52 UTLJ 349
Williams G, ‘Causation in the Law’ (1961) 19 CLJ 62
Witting C, ‘Duty of Care: An Analytical Approach’ (2005) 25 OJLS 33
— — ‘The House That Dr Beever Built: Corrective Justice, Principle and the Law of Negligence’
(2008) 71 MLR 621
x
Wright R, ‘Causation in Tort Law’ (1985) 73 Cal L Rev 1735
— — ‘Actual Causation vs. Probabilistic Linkage: The Bane of Economic Analysis’ (1985) 14
JLS 435
— — ‘The Efficiency Theory of Causation and Responsibility: Unscientific Formalism and False
Semantics’ (1987) 63 Chi-Kent L Rev 553
— — ‘Causation, Responsibility, Risk, Probability, Naked Statistics, and Proof: Pruning the
Bramble Bush by Clarifying the Concepts’ (1988) 73 Iowa L Rev 1001
— — ‘Substantive Corrective Justice’ (1992) 77 Iowa L Rev 625
— — ‘Right, Justice, and Tort Law’ in David G Owen (ed) Philosophical Foundations of Tort Law
(Clarendon Press 1995)
— — ‘Once More into the Bramble Bush: Duty, Causal Contribution, and the Extent of Legal
Responsibility’ (2001) 54 Vand L Rev 1071
— — ‘The Grounds and Extent of Legal Responsibility’ (2003) 40 San Diego L Rev 1425
— — ‘Acts and Omissions as Positive and Negative Causes’ in Neyers, Chamberlain and Pitel
(eds) Emerging Issues in Tort Law (Hart Publishing 2007)
— — ‘Liability for Possible Wrongs: Causation, Statistical Probability, and the Burden of Proof’
(2008) 41 Loy L A L Rev 1295
— — ‘Proving Causation: Probability versus Belief’ in Richard Goldberg (ed.) Perspectives on
Causation (Hart Publishing 2011)
— — ‘The NESS Account of Natural Causation: A Response to Criticisms’ in Richard Goldberg
(ed.) Perspectives on Causation (Hart Publishing 2011)
xi